Reactive Aerosol Jet Printing of Ag Nanoparticles: A New Tool for SERS Substrate Preparation

Abstract

The detection of trace chemicals at low and ultra-low concentrations is critical for applications in environmental monitoring, medical diagnostics, food safety and other fields. Conventional detection techniques often lack the required sensitivity, specificity, or cost-effectiveness, making real-time, in situ analysis challenging. Surface-enhanced Raman spectroscopy (SERS) is a powerful analytical tool, offering improved sensitivity through the enhancement of Raman scattering by plasmonic nanostructures. While noble metals such as Ag and Au are currently the reference choices for SERS substrates, fabrication methods should balance enhancement efficiency, reproducibility and scalability. In this study, we propose a novel approach for SERS substrate fabrication using reactive Aerosol Jet Printing (r-AJP) as an innovative additive manufacturing technique. The r-AJP process enables in-flight Ag seed reduction and nucleation of Ag nanoparticles (NPs) by mixing silver nitrate and ascorbic acid aerosols before deposition, as suggested by computational fluid dynamics (CFD) simulations. The resulting coatings were characterized by X-ray diffraction (XRD) and scanning electron microscopy (SEM) analyses, revealing the formation of nanoporous crystalline Ag agglomerates partially covered by residual matter. The as-prepared SERS substrates exhibited remarkable SERS activity, demonstrating a high enhancement factor (10⁶) for rhodamine (R6G) detection. Our findings highlight the potential of r-AJP as a scalable and cost-effective fabrication strategy for next-generation SERS sensors, paving the way for the development of a new additive manufacturing tool for noble metal material deposition.

1. Introduction

The detection of chemicals such as drugs, pollutants, and trace compounds at ultra-low concentrations is critically important across diverse application fields, including environmental monitoring, medical diagnostics and food safety [1,2,3,4,5,6,7]. These applications usually demand highly sensitive analytical techniques capable of identifying substances in a concentration ranging from the micromolar to nanomolar scale or parts-per-million (ppm) to parts-per-billion (ppb). Conventional detection methods, while effective, often lack sensitivity and specificity or they rely on complicated and high-cost analysis techniques that make those methods inconvenient for real-time and on-site applications. For instance, pollution due to PFAS is emerging as a significant health emergency in many countries worldwide. Due to strict regulations that limit their concentration in water in the parts-per-trillion (ppt) range, highly sophisticated chromatographic analysis techniques are required for their detection [8,9]. As a result, there is an urgent need to develop rapid and cost-effective detection methods for identifying specific compounds at exceptionally low concentration. In this context, conventional analytical techniques such as high-performance liquid chromatography (HPLC), gas chromatography (GC), and mass spectrometry (MS) are widely used for the detection of trace-level contaminants. While these methods are highly accurate, they often require complex sample preparation, expensive instrumentation and extended running times, which limit their applicability in on-site detection scenarios [10]. Surface-enhanced Raman spectroscopy (SERS) has emerged as a powerful tool to overcome these limitations, offering unparalleled molecular fingerprinting capability with high sensitivity [11]. Known since the 1970s [12,13], the SERS effect relies on the amplification of Raman scattering signals when molecules are adsorbed onto or placed near plasmonic nanostructures, typically composed of noble metals. These nanostructures create localized surface plasmon resonances (LSPRs) under specific excitation wavelengths, significantly enhancing the electromagnetic field and thus the Raman scattering signal of nearby analytes. This enhancement factor (10⁶–10¹²) [14], combined with the high sensitivity, chemical selectivity and the potential for real-time, cost-effective measurements, makes SERS a compelling solution for detecting trace chemicals in complex matrices.

The development of effective SERS sensors strictly depends on the design and preparation of suitable substrates that maximize enhancement while ensuring reproducibility and stability. Among the most commonly employed materials, nanostructured noble metals (Ag and Au) are by far the most widely adopted choice, due to their convenient plasmonic properties in the visible to near-infrared range [15]. Beyond noble metals, other materials such as bimetallic alloys, semiconductors, and 2D materials and their composites have gained attention for their potential to combine plasmonic and catalytic effects or to offer additional functional properties [14,15]. In terms of manufacturing, traditionally, SERS substrates were obtained by synthetic methods to fabricate various metal nanostructures and immobilize them onto solid surfaces by deposition techniques. The key advantage of this method is the high degree of control on the shape and size of the metal nanoparticles (NPs) to tune the enhancement factor (EF), but their aggregation on the substrate determines the final SERS capability and the repeatability of analyses. To overcome these limitations, lithography-based techniques were explored to fabricate SERS substrates through well-defined, accurate and periodic micro and nanostructures obtaining high EFs (10⁶–10¹⁰) [16,17]. For instance, techniques such as electron-beam lithography, focused ion beam milling, self-assembly and nanoimprint lithography, and others were demonstrated to be suitable for SERS substrate preparation, as K. Srivastava et al. described in details in their recent review [17]. While these approaches have advanced significantly, they often face limitations in scalability and cost-efficiency. To address these challenges, additive manufacturing (AM) technologies, in particular printing-based methods, have emerged as innovative tools for SERS substrate fabrication. Taking advantage of their low cost, easy scalability and reduced material waste, many additive manufacturing techniques have been recently investigated as promising SERS substrate fabrication methods [18]. For instance, electrochemical additive manufacturing (ECAM) has been proposed to deposit small (20 nm) Ag NPs on a ITO-coated glass to detect Aspirin up to the nanomolar concentration with EF~10⁷ [19]. Through a combination of 3D printing and post-functionalization of the polymeric surface with a thin metal layer or metal NPs, to impart the SER effect, it is possible to produce SERS substrates even with complex geometry [18]. For instance, PLA/Cu 3D-printed composite were easily converted to a Ag-based SERS substrate by galvanic displacement reaction between Cu and Ag [20]. Among additive manufacturing (AM) ink-based techniques, inkjet printing (IJP) has been the most studied for SERS sensor preparation due to its precise control over the quantity of jetted NPs and their uniform distribution, either as a homogeneous layer or in periodic patterns. Some recent remarkable examples are represented by printing Au nanospheres on hydrophobic paper for ultra-low detection (10⁻¹¹ M) of thiram with SERS sensor long-life stability and cost effectivity (<0.01 USD/spot analysis) [21], a mercaptobenzoic acid SERS sensor prepared by Au NPs IJ deposition on a paper substrate with a EF~10⁷ [22] and illegal drugs (methamphetamine) detection in human urine samples by SERS substrates based on Ag NP ink printed on hydrophobic filter paper [23]. Aerosol Jet Printing (AJP) is also standing out as innovative jet-based AM technique in many fields, in particular printed electronics and multi-purpose sensor micro-fabrication [24]. In fact, AJP offers unprecedented advantages for the additive manufacturing fabrication of high-spatial-resolution features (up to ~10 μm) with both 2D and 3D architectures on virtually any kind of substrate, owing to the wide nozzle-to-substrate distance (1–11 mm). However, despite the increasing popularity of AJP in sensors, printed electronics and even energy storage, the application of AJP for SERS fabrication was poorly investigated, and just a few works in the literature proposed SERS substrates prepared by AJP of Au NPs [25,26]. Only recently Ag and Ag/graphene coatings, achieved by AJP on Kapton, were demonstrated as innovative perfluoroalkyl compounds detector with unprecedented sensitivity (up to ~0.4 ppt) for perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) [27]. These works highlight the promise of integrating AJP with nanotechnology to address key challenges in SERS substrate development. However, limitations to industrial applications of AJP for SERS sensors may arise from the need for stable metal NP ink with a long shelf-life, low cost and suitable nanoparticle morphology for SERS activity.

To overcome these limitations and further expand the potential of AJP, inspired by a seminal research on the use of AJP to print combinatory materials [28], we propose herein an innovative approach to print metallic Ag coatings by reactive Aerosol Jet Printing (r-AJP). In fact, although the reactive printing concept has been previously reported for inkjet printing (IJP) [29,30,31,32], a reactive process in Aerosol Jet Printing has not yet been reported. Differently from other reported works, Ag NP reduction and precipitation was initiated during the aerosol droplet flight by the mixing of two aerosol streams. Namely, solutions of AgNO₃ ascorbic acid (AA) were nebulized to produce the Ag⁺ precursor aerosol and reductive aerosol, respectively. The two aerosol streams were in-flight mixed to nucleate Ag seeds before the jetting on the substrate. As a result, a thin coating of metallic Ag NPs was produced on the glass substrate. To quantitatively evaluate the degree of mixing between the two fluid streams a computational fluid dynamics (CFD) analysis was performed [33]. This methodological approach enables a detailed examination of the hydrodynamic interactions and mixing efficiency within the system. The r-AJP of silver is easy and quick compared to the long and tedious synthetic method of producing Ag NPs by chemical precipitation, as well as the formulation of stable Ag NP inks. The as-produced coatings were made of crystalline Ag and a low amount of unreacted and by-product material, as detected by XRD. The r-AJP substrates prepared were thin (1.53 µm for 5 overprinting layers) and characterized by nanoporous aggregates of almost spherical nanoparticles of ~310 nm immersed in a patina matrix of by-product matter. Moreover, owing to their peculiar nanoporous structure, the Ag coating prepared by r-AJP showed remarkable SERS activity to rhodamine (R6G), with high EF (10⁶) for the five-layer substrate, demonstrating that r-AJP of silver could be employed as a new convenient fabrication tool for the preparation of SERS sensors.

2. Materials and Methods

2.1. Reactive Aerosol Jet Printing (r-AJP) System

The r-AJP of Ag was achieved using a self-built apparatus. The system was composed of two main parts: (a) the aerosol nebulization units and (b) the printhead motion system. The nebulization units consisted of 3D-printed tanks equipped with stainless steel mesh nebulizers (5 µm mesh) working at 110 kHz. The nebulization rate was adjusted through a control unit (Tekceleo, Mougins, France, Kit Plug&Spray-T45-MXX NEBx01 IHM). The two aerosol streams were conveyed in PTFE tubes (internal diameter 4 and 2 mm) by a N₂ flux and mixed through a Y polypropylene fit. Sheath N₂ flux was injected into the printhead, and both the sheath and aerosol N₂ flow rate were controlled by mass flow controller (Omega Engineering, Irlam, Manchester, UK, FMA5510A). The mixed aerosol stream was jetted by the printhead through a polyethylene tip of 0.3 mm diameter nozzle. The motion of the printhead was provided by a custom-modified 3D printer (Creality, Shenzhen, China, Ender 3).

2.2. Ag Deposition by r-AJP

In a typical procedure, fresh AgNO₃ (Sigma Aldrich, München, Germany) and l-ascorbic acid (AA, Alfa-Aesar, Heysham, UK) solutions were both prepared at 0.1 M concentration to have a 1:1 molar ratio between Ag ions and AA. The nebulization rate was adjusted to 10% (0.08 mL/min) by the control unit, while the aerosol N₂ conveying the gas stream and the sheath gas stream were adjusted to 40 mL/min and 120 mL/min, respectively. Printing was performed at a 5 mm/s speed while adjusting the layers counts, keeping the nozzle-to-substate distance at 5 mm. Plasma-activated glass slides were employed as the substrate. Oxygen plasma-activation was performed for 1 min in an ATTO plasma cleaner (Diener Electronics, Ebhausen, Germany). During printing the glass slides were placed on an anti-slip silicone pad, heating the heat-bed at 60 °C. For comparison, Ag NPs were synthetized by chemical precipitation via Ag⁺ reduction by trisodium citrate (TSC). Following the Lee–Meisel recipe [34], 40.6 mg of AgNO₃ was added to a flask containing 235 mL of deionized water while a solution of TSC 1% w/w was prepared in another flask by adding 47.6 mg of TSC in 4.7 mL of deionized water. Both solutions were continuously stirred, heated up to boiling point, and mixed. The boiling was then maintained for 20 min, until the color of the solution stabilized. Impurities were removed by multiple centrifugations at 5000 rpm and washing, and the final Ag NP content in the colloidal ink was adjusted to 3% w/w. The colloidal Ag NP ink was printed by AJP using the same process parameters described previously.

2.3. CFD Simulation

In the present study, the evolution of the aerosol species inside the reactive Aerosol Jet Printing system was studied to obtain a qualitative measure of the mixing level of the two streams. To ensure effective mixing of the chemical species, an Eulerian framework was employed [35]. The multiphase flows were described as interpenetrating continua, through the concept of volume fraction ($\alpha$) of the dispersed phase, defined as $\alpha = {\displaystyle \frac{Q_{d }}{\left(Q_{c }+Q_d\right)}}$, where $Q_{d }$ and $Q_{c }$ are the volumetric flow rates of the dispersed and continuous phase, respectively. The continuity, momentum and volume fraction equations were performed by means of ANSYS Fluent version 2021 R2. Specifically, a second-order upwind was employed for the continuity and momentum equations and a QUICK (Quadratic Upstream Interpolation for Convective Kinematics) scheme was used for the volume fraction equation [36]. The modeled geometry was discretized using an unstructured mesh composed of polyhedral elements. Furthermore, a mesh independence study was conducted as part of the simulation process [37]. This involved systematically varying the mesh density to identify the minimum number of cells necessary to obtain results that are independent of further mesh refinement, thereby ensuring the reliability and robustness of the numerical predictions. It was found that a mesh composed of roughly 1 million cells was sufficient to ensure robust results. The turbulence characteristics of the fluid flow were captured using the Shear-Stress Transport (SST) k–ω turbulence model. This model was selected due to its proven capability to accurately represent a wide range of flow conditions, encompassing both laminar and turbulent regimes. The SST k–ω model combines the advantages of both the standard k–ε and k–ω models, enabling it to provide reliable predictions in scenarios where the flow undergoes transition or involves adverse pressure gradients, making it especially suitable for complex engineering applications. To simulate the mixing evolution inside the reactive printer, it was assumed the overall flow of nitrogen (40 mL/min) was equally distributed in each inlet section (20 mL/min), whereas the flow of the aerosol in the two tubes was equal to the nebulization rate (0.08 mL/min).

2.4. Characterization

The microstructure of the Ag NP coatings was analyzed by X-ray diffraction (XRD) in thin-film mode using a PW1830 diffractometer (Philips, Amsterdam, The Netherlands, Kα1Cu = 1.54058 Å). The morphology was observed through a scanning electron microscopy (SEM, EVO 50 EP, Zeiss). Optical profilometry was performed through Sensofar S neox 090 (Sensofar, Terrassa, Spain) in interferometric mode (ePSI algorithm) equipped with a Nikon-DI 20X objective. The silver colloid and the aerosol jet-printed substrates were both characterized by the UV-Vis spectrometer UV-Jasco V570 (JASCO, Cremella, Italy). To acquire and compare the value of the plasmon peak, the spectral range was set between 300 and 800 nm. The as-prepared silver nanoparticles were diluted by one order of magnitude and poured into a cuvette to measure the spectrum. On the other hand, the glass slide carrying the aerosol jet-printed substrate was inserted into a specific part of the UV-Vis machine to perform the spectrum as well. The SERS analyses were performed by the Horiba LabRAM HR800 (Horiba, Kyoto, Japan), employing the 785 nm excitation wavelength, with an integration time of 15 s averaged over 2 accumulations. The laser power was set at 1 mW. Rhodamine 6G in concentrations between 5 × 10⁻⁵ M and 10⁻⁹ M was employed to optimize the printing conditions of the substrates and test their limit of detection. The preparation of the SERS samples consisted of printing an array of Ag substrates as 3 × 3 mm squares and waiting at least 12 h before the SERS analysis. The R6G solution (3 µL) was dropped on each Ag printed square, and once the solution was completely dry by room temperature evaporation, the Raman analyses were performed on at least 3 random points for each Ag substrate. Once the measurements were completed, we analyzed the data calculating the mean and standard deviation value of the spectra collected on the same square to plot and compare the relative value of intensities. Enhancement factor (EF) was calculated at the limit of detection of the SERS substrate, according to previous works [38,39], using the formula EF = (I_SERS/I_NR) × (N_NR/N_SERS) in which I_SERS and I_NR are the intensity of a specific Raman peak in presence and absence of the SERS substrate, respectively, while N_NR and N_SERS are the number of molecules of analyte casted on a bare substrate and the SERS substrate, respectively. The Raman spectrum of R6G was obtained dripping 5 µL of R6G 10⁻⁴ M onto an aluminum sheet. To calculate N_NR and N_SERS, it was assumed that molecules homogeneously distributed on the bare surface as well as were homogeneously adsorbed on the SERS substrate. We normalized the casted volume over the actual surface wet by dripped solution. N_NR and N_SERS were then calculated considering the laser spot area according to the equation $d={\displaystyle \frac{1.22\lambda }{NA}}$ where d is the laser spot diameter, NA the numerical aperture (here 0.75) and laser wavelength λ is 785 nm.

3. Results

3.1. r-AJP System and Modeling

As discussed in the Introduction section, the main advantage of the proposed r-AJP technique is the homogeneous mixing of precursor solutions and confined in-droplet nucleation of Ag nanoparticles during the aerosol flight time. In Figure 1a a schematic representation of the aerosol recombination process is provided, highlighting the aerosol droplets mixing and occurring after the Y fit joint, while Figure 1b,c show the experimental setup employed for the r-AJP process. The nucleation and precipitation of metal NPs in confined droplets is similarly adopted in the water-in-oil or oil-in-water microemulsion precipitation method, with advantages such as nanoparticles narrow size distribution and good growth control by the adoption of specific surfactants and stabilizers [40,41]. As for SERS application, porous, sharp and dendritic Ag structures are well performing in the Raman scattering enhancement, and we intentionally avoided the employment of capping agents, stabilizers and surfactants to promote the confined nucleation and precipitation of silver seeds, limiting the growth step to achieve a nanostructured yet porous coating of Ag nanoparticles. As the reducing agent, l-ascorbic acid (Vitamin C) was employed as a green but efficient reducing agent commonly used in silver nanoparticle synthesis. To favor the quick precipitation of small Ag nanoparticles, AA was used in excess (Ag:AA = 1:1) with respect to the theoretical stoichiometric ratio with Ag⁺ ions (Ag:AA = 2:1) according to the reaction [42]:

2AgNO₃ + C₆H₈O₆ → 2Ag + C₆H₆O₆ + 2HNO₃

(1)

To investigate the probability of in-flight Ag nucleation, namely if silver seeds and nanoparticles were formed before or after aerosol droplets were jetted from the nozzle, the mixing of the two aerosol streams along the tube was simulated. In fact, given the low kinetic constant (0.00042 s⁻¹) [43] and the short residence time (approximately 5 s) in PTFE tube after aerosol recombination zone before entering the printhead unit, mixing emerges as the primary limiting factor for triggering the reaction. Figure 1d,e represent the discretized modeled geometry using an unstructured mesh composed of polyhedral elements. The analysis of volume fraction distribution within the reactive printer provides insights into the degree of segregation between the two fluid streams. As indicated in Figure 1f, red-colored regions denote zones of liquid phase accumulation, signifying poor mixing and high segregation. In contrast, a uniform color distribution suggests effective mixing, implying enhanced radial dispersion and homogeneity of the two streams. Figure 1f demonstrates that, upon entry into the tubes, the denser phase (aerosol) exhibits a tendency to accumulate along the tube walls, indicating a clear phase segregation at the initial stage of the flow. This behavior is likely due to density-driven stratification as the streams first come into contact. However, after the Y-junction, the distribution of the components becomes noticeably more uniform across the cross-section, indicating that mixing is promoted at this point. Although the velocity profile remains laminar, as shown in Figure 1d, the Y-junction induced sufficient perturbations to promote homogeneous mixing. These observations support the conclusion that Ag seeds or nanoparticles are nucleated in-flight by aerosol mixing before being jetted on the substrate. In fact, the mixing of the two bulk solutions resulted in the instantaneous formation of a grayish precipitate of metallic silver due to the quick reduction in Ag⁺ by ascorbic acid oxidation (reduction potential 0.35 V vs. SHE [44]). A few seconds of mixing are indeed enough to induce Ag NP nucleation, similarly to what occurred in mixing of bulk AgNO₃ and AA solutions that instantaneously turned dark and grayish (Video S1). As a result, when printing, grayish matter was deposited on the glass slide (Video S2).

3.2. Characterization of Printed Ag Substrates

XRD analyses were performed on samples prepared by r-AJP adjusting the overprinting to 3, 5 and 10 layers. Results are shown in Figure 2a. Diffraction patterns clearly showed that metallic Ag was deposited, as the peaks at 37.9°, 44.1° and 64.3° correspond to the (111), (200) and (220) planes, respectively, of the face-centered cubic Ag crystals (JCPDS #01-087-0717). Moreover, the very high intensity of the (111) to (200) peak, (I₁₁₁/I₂₀₀ ~16.6 for all the samples) suggest preferential growth along the (111) plane. In terms of SERS activity, the (111) orientation of Ag is favorable as highlighted by previous works [45,46,47]. In fact, as the free energy of the (111) plane is the lowest among the four major planes’ fcc Ag structure, it could adsorb molecules more strongly than the other surfaces, increasing the chemical enhancement of SERS [45]. Notably, clear fingerprints of residual impurities of AA (JCPDS #00-022-1536) or by-products (dehydroascorbic acid) were detected in the 10-layer sample only, where a few low-intensity peaks raised from the background (Figure 2b). Overall, XRD showed that Ag ions were completely reduced and converted to metallic silver, as no residual AgNO₃ was detected, owing to the excess AA employed in this work. Optical profilometry was used to assess the silver coating thickness at five overprinting layers. The height map image (Figure 2c) showed homogeneous and consistent coating height, with average surface roughness (R_a) of 0.189 µm and root mean square (S_q) of 0.298 µm. From the height profile (Figure 2d) the average coating thickness was estimated to be 1.538 µm, and just a minimum overspray effect was clearly visible in the surrounding area of the coating.

SEM analyses showed that for up to five overprinting layers (Figure 3a,b) Ag nanoparticles were precipitated, forming porous and sharp dendritic structures with an irregular shape as neither stabilizing agents nor surfactants were employed, resembling dendritic Ag reported in previous works [48]. On the contrary, at 10 overprinting layers (Figure 3c), almost-compact yet lamellar structures were produced, suggesting that silver ion reduction and nanoparticle growth were partially occurring even after deposition on the substrate or unreacted products had sufficient time to rearrange in more defined crystalline structures. In comparison, a 5 L coating obtained by Aerosol Jet Printing of the colloidal Ag NP ink produced a smooth surface with few globular macro-agglomerates visible on the surface (Figure 3d), as the Ag NPs efficiently packed to form a homogeneous layer during AJP deposition. However, it is widely reported that sharp, dendritic metallic nanostructures are preferable for SERS application as the LSPR enhancement which leads to strong surface-enhanced Raman scattering is typically concentrated only on the curved sharp surfaces of nanostructures containing sharp tips [49,50]. Therefore, smooth coatings of Ag NPs are not expected to result in an appreciable SERS effect. EDX mapping for the 5-layer sample (Figure 3e) showed that Ag was homogeneously distributed along with carbon and nitrogen, probably coming from a trace of residual silver nitrate, confirming that metallic silver precipitated along with a by-product matter as suggested by XRD results. This was particularly evident up to 5 layers, as shown in (Figure 3f,g), where the silver nanoparticle network was partially covered by a homogeneous coating and immersed in a patina matrix. After washing the samples, part of the printed material, including silver nanoparticles, was detached and removed from the substrate, but overall the surface coverage resembled the one before washing (Figure 3h,i). The remaining porous matrix comprised a 3D network of almost-round-shaped silver nanoparticles of 310 ± 75 nm, and residual AA was not detected.

3.3. SERS Behavior of Ag Substrates

UV-Vis (Ultraviolet–Visible) spectroscopy is crucial for detecting plasmonic peaks in SERS substrates because it provides direct information about the LSPR of the nanostructured material. The spectra resulting from the UV-Vis experiments are reported in Figure 4. The plasmon resonance peak of the colloidal Ag NPs dispersion was registered at 416 nm (Figure 4a), consistent with the expected LSPR of well dispersed spherical Ag nanoparticles in aqueous medium [51]. In contrast, the spectrum collected from the r-AJP substrate exhibits a broader LSPR band, centered at 430 nm (Figure 4b). This red-shift and peak broadening can be attributed to the dense packing of nanoparticles on the substrate and the resulting near-field electromagnetic coupling between adjacent particles. Such dipolar interactions modify the resonance condition, leading to a lower-energy plasmon mode and consequently a red-shift in the LSPR band [52]. Furthermore, the change in the dielectric environment, from a homogenous aqueous medium in the colloidal sample to a heterogeneous interface involving glass, air and any residual organic layers in the substrate, also plays a role in modulating the observed LSPR [53]. These observations confirm that substrate-based assembly leads to distinct plasmonic behavior compared to colloidal systems, driven primarily by interparticle coupling and environmental refractive index [54]. Moreover, a wider plasmonic resonance broadens the spectral range in which electromagnetic enhancement can occur, thus increasing the efficiency of the substrate across multiple excitation wavelengths. This is particularly relevant for analytical applications, where laser sources may vary depending on instrumental configurations or the spectroscopic needs of specific analytes [55,56].

The SERS substrates prepared by standard Aerosol Jet Printing of colloidal silver nanoparticles did not show good performances in Raman scattering enhancement. In fact, it was not possible to observe rhodamine (R6G) signals even when a solution of high concentration (10⁻⁵ M) was drop-casted on the printed substrate. The only weak peaks that are present both in the blank and the R6G spectra are probably related to some unreacted species present on the surface of the substrate (Figure 5a). According to SEM analysis this behavior could be ascribed to the compact and smooth silver coating that did not provide a sufficient LSPR effect, resulting in negligible Raman signal intensity improvement. On the other hand, the silver substrate prepared by r-AJP method showed good SERS effect. In particular, we investigated the effect of overprinting layers on the overall rhodamine detection at an R6G concentration of 5 × 10⁻⁵ M. The spectra were collected on three random points of the substrate, avoiding the edges. From the corresponding Raman spectra (Figure 5b), it resulted that the 5-layer substrate showed the highest relative intensity, followed by the 3- and then the 10-layer SERS substrate. In fact, considering the relative intensities of the 614, 1363 and 1512 cm⁻¹ peaks of R6G, respectively (Figure 5c), it is clear that the 5-layer substrate outperformed the other samples, almost doubling the relative intensity with respect to the 3-layer substrate. In general, the SERS signal is greatly affected by many parameters including SERS active layer thickness, as reported by Lee et al. [57], which demonstrated that as the thickness of a SERS substrate increases, the localized plasmons become less effective, with a significant drop in SERS intensity. Due to its best SERS activity, five overprinting layers was defined as the optimal condition for SERS substrate preparation and the assessment of the R6G detection limit. The R6G limit of detection of the five-layer substrate was registered at 10⁻⁸ M (Figure 5d), similarly to previous works employing Ag-based nanostructures deposited by Aerosol Jet Printing of a commercial Ag NP ink [27], PVD and annealing of Ag thin film [45] or drop-casting complex hierarchical shape silver nanoparticles [58]. In contrast, the r-AJP method we here propose is able to quickly produce SERS substrates at a speed of 4.3 s/mm²/layer. However, at such a low concentration, the reproducibility of the measurements decreased, as is clear from the standard deviation in Figure 5d. The enhancement factor (EF) was estimated to quantify the effectiveness of the Raman signal improvement when the analyte is located near a SERS-active surface compared to its signal under normal conditions. It is noteworthy that the EF is highly dependent on the specific characteristic of the analyte, for example, its molecular structure, adsorption properties and its ability to interact with the SERS-active substrate [59]. Hence, the EF is not a universal constant, but rather specific to each analyte–substrate combination [60]. We selected the 1363 cm⁻¹ SERS band, related to the C-C vibrational mode, following Canamares et al. work [61]. This band was chosen because it is the most stable under different experimental conditions, has generally high intensity, exhibits small changes upon ionization effects and molecular state changes, and it is well separated from other peaks in the Raman spectrum. The EF was calculated to be 10⁶. This value is in good agreement with previous works employing glass-supported flower-like Ag nanoparticles showing an R6G detection limit up to 10⁻⁷ M and EF 10⁴–10⁵ [62,63]. However, it is noteworthy to specify that the EF and lowest detection limit values reported in the literature are very scattering, but in some cases ultra-low R6G concentration detection, even up to 10⁻²⁰ M, has been demonstrated [64]. However, this ultra-high sensitivity is usually achieved through a complex combination of a very fine control over the shape and dimensions of Ag NPs [65], the synergic effect of supporting materials as MOF and SiO₂ [64,66] or the effect of rough substrates [67]. As our focus was the demonstration of the r-AJP technique as a new fabrication tool, we did not pursued the optimization and tuning of the as-obtained Ag nanoparticles as it was out of the scope of this study, and the EF and R6G detection limits fall within the average of the reported values in the literature, thereby demonstrating the viability of r-AJP for producing well performing SERS sensors.

In summary, in this work we introduced for the first time r-AJP as a new technique for the easy deposition of Ag nanostructured coatings that could be conveniently used for the fabrication of SERS sensors for R6G detection at low concentration. In this work we intentionally used very simple solutions of Ag⁺ precursor (AgNO₃) and green reducing agent (l-ascorbic acid) without employing surfactants, complexing or capping agents to tune silver seed nucleation and promote specific crystal growth. We also showed that the as-prepared silver coatings included residual unreacted matter and by-products. Despite their presence, which may interfere with the analyte affecting the SERS behavior, good SERS activity was demonstrated. However, strategies for the consolidation and substrate sticking of the silver nanoporous layer should be investigated aiming to allow washing and removal of reaction residues without affecting the Ag nanoporous coating. Moreover, further limitations may arise from the limited aerosol mixing time that may narrow the range of materials that can be deposited to noble metal only and/or making difficult the synthesis of specific nanostructures with well-controlled geometry, unless particular surfactants or additives are added in the precursor aerosol solutions.

4. Conclusions

In this work, we proposed reactive Aerosol Jet Printing (r-AJP) as a new additive manufacturing tool for the deposition of Ag nanoporous coatings. In general, the r-AJP process is simple and, in principle, it could be employed for the deposition of any material produced through quick reactions (~s) occurring by two or more aerosol streams mixing. For instance, we demonstrated the viability of the r-AJP technique for the deposition of nanoporous silver coating on a glass slide, owing to the fast precipitation of Ag nanoparticles by ascorbic acid oxidation. CFD simulation suggested that homogeneous mixing of the two aerosols occurred after the recombination zone (Y fit joint) promoting Ag seed formation. The as-produced Ag coatings were formed by silver crystalline nanoparticles aggregates and unreacted material, due to the excess ascorbic acid employed to ensure quick and complete Ag⁺ reduction. The r-AJP substrates were used as SERS sensors for rhodamine detection down to the low concentration of 10⁻⁸ M. It was found that SERS activity to rhodamine was dependent on overprinting layers number, with the best sensitivity achieved for five layers, showing a Raman enhancement factor of 10⁶. We believe this work may lay the foundation for future advancements and applications of r-AJP, while also serving as a reference for further research into reactive-based additive manufacturing techniques.