Journal Archive(XML)

Seed-mediated growth separates the nucleation and growth phases, providing enhanced control over size, shape, and structural uniformity. This method enables the systematic synthesis of various anisotropic architectures (Fig. 1) such as rods, cubes, bipyramids, and nanostars by controlling three key aspects: facet-surfactant interactions (Section 2.1.1), reducing agent selection and kinetics (Section 2.1.2), and solution conditions, including pH and environmental factors (Section 2.1.3). Each of these parameters offers distinct pathways for directing the growth and tailoring the resulting plasmonic properties.

Fig. 1. 
Representative morphologies of plasmonic gold nanostructures synthesized via seed-mediated growth, illustrating diverse anisotropic architectures. (a) Nanorods obtained by aspect ratio control, adapted from Ref. [21]. (b) Nanocubes and octahedra illustrating facet-controlled growth, adapted from Ref. [22]. (c) Bipyramids formed by regulating growth at high-energy sites, adapted from Ref. [23]. (d) Nanotriangles resulting from kinetically controlled planar growth, adapted from Ref. [24]. (e) Nanostars produced through surfactant-directed branched growth, adapted from Ref. [25]. (f) Nanoplates synthesized by tuning thickness during 2D growth, adapted from Ref. [26].

Although solution-phase synthesis enables the formation of plasmonic nanostructures with diverse morphologies, the random distribution and aggregation of colloidal particles often limit their practical use in sensing applications. In particular, drop-cast or dried colloidal systems frequently suffer from heterogeneous analyte adsorption, uneven particle distribution, and coffee ring effects, all of which undermine spatial uniformity and signal reproducibility [43-45].

To overcome these limitations, guided fabrication strategies employ structural assistance, patterned templates, or substrate-anchored growth environments to direct the nucleation, assembly, and growth of plasmonic nanoarchitectures. These approaches improve spatial precision and enhance SERS performance across sensing areas. Representative strategies include nanoporous materials, mesoporous silica frameworks, block copolymer (BCP) self-assembly, nanoimprint lithography, and solution-phase assembly. Each strategy offers distinct advantages in terms of structural control, scalability, and compatibility with different substrates.

2.2.1 Nanoporous Anodic Aluminum Oxide for Guided Plasmonic Nanomaterials

A classic example of confined-geometry fabrication is anodic aluminum oxide (AAO) templating, in which vertically aligned cylindrical pores serve as microreactors. In conventional AAO, the pore shape and spacing are coupled with the anodization parameters, limiting independent control over the geometry, periodicity, and orientation [46]. These confined channels enable the localized reduction of metal precursors, leading to the formation of one-dimensional nanostructures such as nanowires and nanotubes. The resulting morphology is governed by the deposition parameters and surface diffusion dynamics during growth. Moreover, annealing at elevated temperatures promoted Ostwald ripening, which narrowed the size distribution and enhanced structural uniformity throughout the array.

To overcome these geometrical constraints, Xu et al. [47] recently combined nanoimprinting and anodization to create programmable AAO templates that went beyond the usual circular pores, allowing stars, crosses, triangles, and other pore shapes with independently tunable spacings and array arrangements in both the lateral and vertical directions (Fig. 2 (a)). This strategy enables the free design of pore geometry, periodicity, and orientation. These templates are compatible with physical vapor deposition and electrochemical filling, allowing the fabrication of silver nanoparticle (AgNP) arrays, gold nanowires, and other plasmonic architectures with tailored shapes for high-performance SERS substrates.

Fig. 2. 
Fabrication of guided plasmonic nanomaterials using diverse templating strategies. (a) Programmable AAO templates with shape-designable pores, adapted from Ref. [47]. (b) Mesoporous silica and monolayer templates with tunable pore chemistry for confined growth, adapted from Refs. [48,49]. (c) BCP-templated freestanding gold network formed by selective polymer removal, adapted from Ref. [51]. (d) Nanoimprinted silver nanorod arrays with built-in nanoscale gaps for biofilm metabolite sensing, adapted from Ref. [53]. (e) SiO₂-tipped gold nanorod plasmonic cavity membranes enabling analyte enrichment and trapping, adapted from Ref. [57].

Paper-based SERS substrates have emerged as promising platforms because of their flexibility, low cost, capillarity-driven fluid handling, and suitability for field-deployable diagnostics. The direct growth of plasmonic nanoarchitectures on cellulose paper or fiber-based scaffolds eliminates the need for complex transfer or printing steps, significantly improving reproducibility and scalability.

One notable example is a 3D plasmonic coral nanoarchitecture (PCN) paper developed for label-free human urine sensing integrated with a deep learning model for cancer screening [62]. In this paper-type SERS platform, gold nanostructures were grown on cellulose acetate (CA) paper through a seed-mediated reduction process involving HAuCl₄and a mild reducing agent. Importantly, a hierarchical coral-like morphology emerged because of interparticle attachment mechanisms, in which small gold nanocrystallites nucleated and then aggregated in a directional and cooperative manner (Fig. 3 (a)). A hierarchical coral-like architecture composed of multiscale branches and interconnected nanopores was formed during the synthesis of the gold nanostructures, facilitating abundant hotspot formation (Fig. 3 (b)). The acquired SERS spectra achieved a classification accuracy exceeding 95% in differentiating normal, prostate cancer, and pancreatic cancer urine samples using a convolutional neural network (CNN)-based deep learning approach. Furthermore, Kim et al. [63] fabricated a 3D plasmonic gold nanopocket (3D-PGNP) structure on CA filter paper, enabling its direct integration with syringe filter devices for on-site analysis. Gold nanoarchitectures were synthesized directly on a CA filter paper through a spontaneous nucleation process involving a gold precursor and O-methylhydroxylamine hydrochloride (OMH) as the reducing agent. In contrast to conventional shape-controlled growth, gold ions are reduced and deposited uniformly across the substrate without preferential affinity for specific crystallographic facets. This uncontrolled deposition results in the formation of spherical polyhedral gold particles with an average diameter of 176 ± 26 nm, which are randomly distributed throughout the cellulose fiber matrix (Fig. 3 (c)). The spatial arrangement of these particles generated microscale voids, referred to as nanopockets, situated between adjacent nanostructures. These nanopockets played a dual functional role. First, they act as physical traps for microplastics (e.g., polystyrene and polyethylene) during syringe-based filtration. Second, the confined plasmonic cavities enhance the local electromagnetic field for SERS enhancement. This dual function significantly improves sensitivity and analyte capture efficiency, with a limit of detection (LOD) for polyethylene as low as 5.3 μg/mL (Fig. 3 (d)). These two approaches illustrate how distinct nanoarchitecture formation mechanisms can be strategically employed to optimize SERS performance and tailor the sensing platform to specific analytical applications. Moreover, Kim et al. [64] developed a label-free cellulose-based SERS biosensor for the early diagnosis of subarachnoid hemorrhage (SAH)-induced complications using human cerebrospinal fluid (CSF) samples. To enhance the nanoparticle-mediated LSPR effect, a pH-controlled functionalization step was incorporated into the successive ionic layer adsorption and reaction (SILAR) process, which enabled the dense and uniform growth of crystalline AuNPs on the paper substrate. In the process of SILAR synthesis, the cellulose surface was initially treated with polyethylenimine to form a positively charged interface, which promotes the electrostatic attraction of negatively charged Au⁺ ions during immersion in a HAuCl₄ solution. Subsequent immersion in a reducing agent solution triggered the in situ reduction of adsorbed Au ions into metallic nanoparticles. Repeating this cycle allowed for the controlled layering and growth of gold nanoarchitectures. Notably, by fine-tuning the pH of the precursor and reducing solutions, the nucleation rate, ion adsorption efficiency, and crystal uniformity were optimized, leading to a significantly enhanced SERS performance through improved hotspot density and LSPR uniformity. Clinically, the SERS biosensor demonstrated a diagnostic accuracy of more than 87% for classifying CSF samples from patients with cerebral vasospasm and hydrocephalus, with a signal variability of 19.3%.

Fig. 3. 
Paper-based plasmonic sensing platforms. (a) Conceptual depiction of the stepwise formation of 3D-PCN structures on CA substrates. (b) Digital image showing the fabricated 3D-PCN structure on the CA paper platform. The inset shows SEM image of the 3D-PCN (scale bar: 500 nm). Results of (a) and (b) adapted from Ref. [62]. (c) Illustration of a paper-type SERS platform for microplastic analysis, highlighting PS particle entrapment within the nanopocketed network. A high-resolution SEM micrograph shows microplastics embedded in the porous 3D-PGNP (scale bar: 500 nm). (d) Standard curve for polystyrene SERS signals measured on 3D-PGNP. Results of (c) and (d) adapted from Ref. [63].

Plasmonic needle-based SERS platforms offer a transformative approach for minimally invasive diagnostics, enabling the direct molecular analysis of tissues or biofluids with minimal patient discomfort [65,66]. By growing SERS-active nanoarchitectures directly onto needle surfaces, these platforms facilitate real-time signal acquisition, while maintaining structural integrity and reproducibility.

In a recent study, a plasmonic needle endoscopy system was developed by directly functionalizing the distal tip of a medical-grade endoscopic needle with 3D polyhedral gold nanostructures, thereby enabling real-time label-free in vivo SERS analysis of colorectal tissue during endoscopic procedures [67]. In this approach, gold nanostructures are directly synthesized on the distal tip of a stainless-steel endoscopic needle via polymer-mediated surface reduction. Initially, the needle surface was functionalized with dopamine, which is rich in amine and hydroxyl groups, serving as both a stabilizing interface and mildly reducing environment. Upon immersion in a HAuCl₄solution, the surface-bound functional groups facilitated the localized nucleation and growth of polyhedral gold nanocrystals. As shown in Fig. 4 (a), the fabricated plasmonic needle sensor (PNS) presents a stark contrast in morphology compared to the unmodified stainless-steel needle (SN). Notably, the high-density and uniformly distributed polyhedral gold nanoarchitectures in the PNS ensured high SERS activity (right panel in Fig. 4 (a)). Comparative SERS analysis of colorectal cancer (CRC) tissue and adjacent mucus revealed similar Raman signatures (Fig. 4 (b)), confirming that colonic mucus can serve as a reliable surrogate for tumor tissue. Recently, an SERS-based MN biosensing platform was developed for in situ and highly sensitive detection of tyrosinase (TYR), a critical enzymatic biomarker involved in melanin biosynthesis that is widely recognized for its diagnostic relevance in melanoma [68]. The SERS platform features a minimally invasive MN array functionalized with dual-purpose nanoparticle systems designed for target recognition and signal enhancement. To fabricate the sensing interface, the MN surface was chemically modified to support the robust immobilization of dopamine-functionalized AuNPs (DA-AuNPs). The catechol moieties present in dopamine play a dual role: they serve as mild reducing agents that facilitate the in situ nucleation and growth of gold nanoparticles, and act as strong adhesive ligands via the chelation of surface metal ions on the MN substrate. This resulted in a stable and uniform plasmonic coating that was precisely localized on the MN surface (Fig. 4 (c)). Upon insertion into the skin, endogenous TYR oxidizes dopamine to dopamine quinone, which reduces the binding efficiency of the SERS probes, resulting in quantifiable changes in the Raman signal intensity. The biosensor exhibited a broad linear detection range of 0.05 to 200 U/mL for TYR, with excellent anti-interference performance.

Fig. 4. 
Needle-type SERS sensing platforms. (a) Photographic image of the unmodified stainless-steel needle (SN) and the PNS, with a SEM image revealing the nanostructured surface of the PNS (scale bar: 200 nm). (b) Representative SERS spectra distinguishing between intestinal mucus from healthy and tumor-bearing mice. Results of (a) and (b) adapted from Ref. [67]. (c) Schematic workflow showing the fabrication of the SERS-integrated microneedle and its operational principle for in situ TYR detection via minimally invasive sampling, adapted from Ref. [68].

Together, these two platforms demonstrate the versatility of direct surface growth strategies for fabricating high-performance needle-based SERS biosensors. Although the endoscopic needle system utilizes polymer-mediated metal growth on rigid metallic substrates for tissue-level diagnostics, the MN array utilizes catechol-driven gold deposition on flexible polymer matrices to enable minimally invasive molecular sensing. These approaches highlight how surface chemistry and growth conditions can be tailored to specific device geometries and diagnostic targets, thereby broadening the scope of plasmonic SERS technologies for clinical applications.

To address the demand for parallelized molecular detection, direct-growth plasmonic substrates have been integrated into standard multiwell plate formats [69]. Direct in-well growth enables the controlled formation of hotspots and robust physical entrapment of analytes within confined plasmonic fields, thereby improving both reproducibility and detection sensitivity.

As a practical implementation of the direct-growth strategy, a 3D evolutionary gold nanoarchitecture (3D-EGN) was directly synthesized within a standard 96-well plate, facilitating whole urine-based multiplex cancer diagnosis and comprehensive metabolite profiling via SERS analysis (Fig. 5 (a)) [70]. To enable high-throughput liquid-phase sensing, a standard 96-well polystyrene plate was used as the substrate for the direct synthesis of 3D-EGN (Fig. 5 (b)). As shown in Fig. 5 (c), the synthesis was carried out directly on a 96-well plate via a three-step process: (i) formation of a nanoporous gold sponge (AuS) through in situ nucleation using a gold precursor and hydroxylamine in ethanol, (ii) attachment of AuNPs onto the AuS surface, and (iii) conformal lamination of a thin gold layer over the structure. In ethanol-rich environments, the nonselective surface adsorption and interaction of methyl groups with the gold surface attenuated directional growth, leading to low-surface-energy with reduced branch lengths and improved uniformity. These 3D nanoarchitectures created densely packed plasmonic hotspots that enabled direct metabolite profiling of untreated clinical urine samples. By utilizing CNN-based machine learning, the well-plate SERS platform successfully classified healthy and cancerous samples such as the prostate, lung, and bladder, achieving over 90% diagnostic accuracy.

Fig. 5. 
Well-plate integrated plasmonic sensing platforms. (a) Schematic illustration of the 3D-EGN platform designed for label-free human urine sensing and multiplex cancer diagnosis. (b) Photograph of the 3D-EGN-integrated 96-well plate. (c) Stepwise schematic of the 3D-EGN fabrication process: in situ formation of nanoporous AuS, attachment of AuNPs, and conformal lamination of a thin Au layer. Results of (a–c) adapted from Ref. [70]. (d) Mechanism of SERS enhancement via the PME effect. (e) SERS spectra obtained from PME structures across varying DNA concentrations, demonstrating sensitivity. (f) Calibration curves derived from concentration-dependent SERS intensities, enabling quantitative nucleotide analysis. Results of (d–f) adapted from Ref. [71].

In a complementary study, the plasmonic molecular entrapment (PME) technique was applied to the detection and quantification of methylated DNA as a key epigenetic biomarker in early stage cancer diagnosis [71]. Similar to 3D-EGN, this PME method utilizes a well-plate-grown AuNP substrate that physically captures DNA strands within plasmonic hotspots (Fig. 5 (d)). The PME system produced precise and consistent SERS signals for DNA sequences, with distinct peak patterns corresponding to adenine, guanine, thymine, and cytosine. The Raman intensity showed a strong linear correlation with the DNA concentration, as summarized in the calibration curves in Figs. 5 (e) and 5 (f). These results highlight the capability of well-plate-integrated plasmonic platforms to perform high-throughput, quantitative, and label-free biomolecular assays with excellent sensitivity and reproducibility. By enabling simultaneous analysis across multiple wells, such systems are ideally suited for scalable clinical diagnostics and screening applications, particularly in scenarios requiring rapid assessment of trace biomarkers in complex biological samples.

Hydrogel film-based wearable SERS platforms have garnered growing interest for noninvasive, real-time biofluid monitoring owing to their conformal skin contact, intrinsic moisture absorption, and compatibility with flexible formats [72,73]. These platforms utilize hydrogels as both sweat-collection interfaces and SERS-active substrates by embedding plasmonic nanomaterials into the hydrogel matrix. This design enables analyte enrichment and stable hotspot formation in a soft, biocompatible form ideal for continuous skin-mounted sensing.

Goh et al. [74] presented a wearable SERS patch that synergistically combined the unique properties of a silk fibroin film (SFF) and silver nanowires (AgNWs) for efficient molecular sensing in sweat. The SFF serves as a moisture-permeable interface, whereas the embedded AgNWs provide high-density plasmonic hotspots (Fig. 6 (a)). This architecture permits direct laser excitation through the dermal patch layer, enabling in situ detection of 2-fluoro-methamphetamine (2-FMA) using a methamphetamine analog. The wearable SERS platform achieved a detection limit of 50 ng/cm² and maintained linear signal response between 2.5 and 50 ng/cm² (Fig. 6 (b)). Notably, 2-FMA spiked in artificial sweat was detected on cadaver skin after patch application without removal, thereby expanding its potential for noninvasive forensic and clinical use.

Fig. 6. 
Hydrogel film-based wearable SERS platforms. (a) Schematic overview of the wearable SERS patch enabling direct molecular detection from sweat. (b) SERS spectra illustrating detection capability for 2-FMA through the patch. Results of (a) and (b) adapted from Ref. [74]. (c) Diagram showing the synthetic pathway for HAP hydrogel formation and its use in clenbuterol sensing. (d) SERS spectral responses from the HAP patch at different clenbuterol concentrations. Results of (c) and (d) adapted from Ref. [75].

Expanding on this concept, a hydroxyethyl cellulose-based polyacrylamide (HAP) hydrogel SERS sensor was designed to achieve ultrasensitive detection of β-adrenergic stimulants, such as clenbuterol, in sweat [75]. As illustrated in Fig. 6 (c), hydroxyethyl cellulose (HEC) was used as the structural backbone for embedding AgNPs, which were synthesized in situ via reduction of AgNO₃and stabilized with disodium citrate. The resulting HEC-AgNP colloid was photocrosslinked with acrylamide to form a flexible skin-adherent hydrogel. The porous mesh structure and hydrophilic nature of the hydrogel enabled the rapid absorption of sweat, promoting the localized enrichment of analytes near the plasmonic hotspots. The wearable sensor demonstrated a rapid response time of under 10 s and achieved a low detection limit of 9.4 × 10⁻⁹ g/L for clenbuterol, with a linear detection range from 10⁻³ to 10⁻⁸ g/L (Fig. 6 (d)). The abundant hydroxyl and carboxyl groups in the HAP hydrogel matrix not only improved the mechanical strength and sweat adsorption but also enhanced skin adhesion through hydrogen bonding, ensuring stable signal acquisition during movement.