Direct printing of gold nanospheres from colloidal solutions by pyro-electrohydrodynamic jet allows hypersensitive SERS sensing

: Pyro-electrohydrodynamic jet printing (p-jet) has been used to fabricate a nanostructured plasmonic sensor for SERS spectroscopy. The p-jet approach allowed us to have an assembly of nanoparticles with suitable geometry and size, which resulted in a sensing surface with intense SERS activity and a rather homogeneous response. Raman imaging measurements highlighted strong Enhancement Factors across the sensing area, exceeding those of the pristine colloidal solution by almost two orders of magnitude. The intense plasmonic effect was further demonstrated by the spectroscopic recognition of a metal-catalyzed dimerization process triggered to completeness at the metal surface. The results presented herein demonstrate the usefulness of the proposed SERS sensor for hypersensitive molecular analysis. and response homogeneity by confocal Raman spectroscopy and Raman imaging. The


Introduction
Surface Enhanced Raman Scattering (SERS) is a plasmonic process by which the signals arising from the inelastic scattering of a molecule are enhanced by several orders of magnitude, (typically 10 5 -10 9 ) resulting in drastically improved sensitivity of Raman spectroscopy measurements. The effect occurs when the target molecule is adsorbed on, or in close proximity to a metal surface with nanoscale morphology. In recent years, SERS has emerged as one of the leading techniques for molecular analysis both in fundamental studies and in an applicationoriented perspective. Remarkable results have been recently reported in such diverse fields as the analysis of trace contaminants [1,2], catalysis [3,4], imaging and nanomedicine for diagnostic and therapeutic applications [5][6][7].
Several strategies have been developed to realize SERS sensors, the most effective being: i) synthesis of individual metal-nanoparticles in the form of colloidal solutions; ii) deposition of colloids onto suitably treated substrates; iii) fabrication of patterned nanostructures by electron beam lithography (EBL). Each of these methods has advantages and drawbacks. Colloid solutions are stable and easy to handle. They represent the main platform for biosensing and provide a highly uniform response, which is a major advantage in quantitative analysis. The signal intensity, however, may change significantly with time if kinetic effects control the adsorption process and/or in case of aggregation and consequent precipitation of the nanoobjects. Furthermore, dilution may appreciably reduce the sensitivity. Planar structures obtained by colloid deposition (drop-casting) are, in principle, the most sensitive platform owing to the formation of a large number of hot-spots (1 -5 nm gaps at inter-particle junctions), which greatly enhance the plasmonic effect [8]. However, uniform coverage of the substrate is difficult to achieve and most systems give rise to the so-called coffee-ring effect. The nanoparticles tend to separate in the boundary region of the dried drop forming a characteristic pattern, which is detrimental not only for the consistency of the response, but also for the quality of the generated signals (fluorescence is frequently observed in areas with a high deposition density). These effects may severely limit the usefulness of planar substrates, especially in quantitative analysis.
Conversely, EBL nanostructures are highly reproducible and essentially defect-free, thus providing very homogeneous response. However, their cost is conspicuous in comparison to other platforms and by this route it is difficult, to reproduce hot-spot morphologies, which tends to reduce the sensitivity.
In the context of the drop-casting strategy, the ability to confine the nanoparticle deposition within micrometric areas presents distinct advantages. The increased nanoparticle density largely enhances SERS sensitivity when the transfer of the analyte to the sensing surface is performed by immersion (dip-transfer). Drop-transfer can be advantageous whenever it is possible to realize a targeted deposition of microdroplets. Another situation in which confined nanoparticle deposition can be highly advantageous is when the sampling volume is very limited. This occurs, for instance, when SERS detection is combined to a preliminary separation step via GC/HPLC, which is becoming a viable route in proteomic/metabolomic research [9,10].
In these applications, micro/nanoliters of the analyte can be mixed with comparable volumes of the sensing colloid, followed by confined deposition of the resulting solution. The quality and intensity of the SERS signals can also be improved when the sensing area has a surface comparable to the laser spot. A typical objective lens used for semi-macro sampling (0.5´, NA = 0.02) produces -with a 632 nm laser line -a beam diameter at focus of around 50 µm, depending on the specific setup. Thus, a laser spot focused at the center of a sensing area limited to less than 100 µm would cover a substantial extent of the nanoparticle deposition, and the SERS intensity sampled in this way would be more representative of the signal distribution compared to a deposition spread over a much larger surface.
In this paper, we propose a method for fabricating a SERS sensor characterized by a micrometric sensing surface having carefully controlled shape and size in the range 50 -100 µm. This nanostructured plasmonic sensor is based on a colloidal solution of gold nanospheres synthesized by seed-mediated growth 25 and characterized by uniform shape and narrow sizedistribution. We used here a non-conventional technique that we call p-jet, that stands for pyroelectrohydrodynamic jet [11], in order to achieve the confined sensing surface by direct printing of the colloidal solution. Compared to more conventional approaches, such as microcontact printing [12], photolithography [13], nanoimprint [14], EBL [15], focused ion beam [16] and direct laser writing [17], p-jet is free from mold-based procedures and does not require the use of expensive equipment. The p-jet uses high electric fields generated pyroelectrically by lithium niobate (LN) and lithium tantalate (LT) crystals [18]. Inkjet printing by the use of electric fields generated by high voltage generators has been developed for different applications [19,20]. The p-jet used here has the additional advantage of being electrode-free and nozzle-free, thus allowing us to avoid external high voltage generators and nozzle-clogging drawbacks [21]. We have already demonstrated the success of p-jet for a wide variety of applications which include printing fine patterns [11,22], patterning tiny fibres [23], accumulation of biomolecules [23], formation of disperse sprayed droplets [24], manipulation of soft matter [25][26][27][28], just to cite some.
Here the p-jet is used for the first time for direct writing of metal nanoparticles from colloidal solution. A complete characterization of the AuNPs is reported, by transmission electron microscopy (TEM), statistical image analysis (SIA) and UV/Vis spectroscopy. Furthermore, we evaluated the SERS performances of the nanoparticle assembly in terms of Enhancement Factor (EF) and response homogeneity by confocal Raman spectroscopy and Raman imaging. The results demonstrate clearly how the proposed method is highly promising in nanotechnology research, as it enables the precise jetting of AuNPs for SERS detection.

Materials
Trisodium citrate (TC), gold(III) chloride trihydrate (HAuCl4 · 3H2O), ascorbic acid (AA), sodium borohydride (NaBH4), Cetyltrimethylammonium bromide (CTAB), ethanol and pmercaptoaniline (pMA) were purchased from Sigma-Aldrich and used as received. Milli-Q water was used for the preparation of gold nanoparticles. The LN crystals were bought from Crystal Technology Inc. in the form of both sides polished 500 μm thick c-cut 3-in. wafers and were cut into square samples 2 × 2 cm 2 sized by a precision diamond saw. The deposition slide was a coverslip, bought from Sigma Aldrich, consisting of an untreated glass substrate 24 × 60 mm sized and 0.2 mm thick.

Pyroelectric effect
The c-cut LN is a ferroelectric crystal and, at room temperature, has a spontaneous polarization Ps conventionally oriented from the so-called c-face to the c+ face. At equilibrium, i.e. when there is no temperature variation, the polarization charge of the LN crystal is completely screened by external screening charges on the crystal surface and there is no electric field. When thermally stimulated, the magnitude of Ps changes and, consequently, a transient electrostatic state appears with uncompensated screening charges on the crystal surface [18]. This phenomenon generates a high electric field that originates from the hot point of the crystal surface. This electric field can be exploited for a wide variety of applications ranging from biological to soft matter manipulation [23,[29][30][31].  We deposited a mother drop (0.2 µL) of the colloidal solution onto a round shaped tip by standard manual pipetting. The p-jet system is made of different elements mounted on threeaxes precision translation stages, aligned vertically from top to bottom. The thermal source consists of a tungsten wire, which heats locally the LN crystal sample by Joule effect. The red rectangle in Figure 1(b) indicates schematically this hot point of the crystal, which corresponds to the area in which the pyroelectric effect occurs. The grey lines in Figure 1(b) show schematically the distribution of the electric field generated across the hot point of the crystal.

P-jet printing
We mount the deposition slide just under the crystal and we deposit the spots of the colloidal solution on the opposite side. The distance between the mother drop and the deposition surface is around 80 m. A conventional optical path, consisting of a collimated LED, an optical microscope objective (10´) and a high-speed CMOS camera (Motion Pro Y3-S1, pixel size of 10.85 μm2), permit us to have a lateral view of the jetting events during the activation of the pyroelectric effect, as shown in Figure 1(a).
The electric field generated pyroelectrically by the LN crystal induces charge accumulation on the free surface of the mother drop by the electrostatic effect. The density of charge is high enough to induce a well-known Coulomb repulsion effect [32,33], which deforms the meniscus of the mother drop into the so-called Taylor cone. When the electric field exceeds the surface tension of the liquid, tiny droplets are ejected from the apex of the meniscus [34]. The daughter droplets generated by this p-jet system have volumes in the range of 80 pL. After the first spot we translate the deposition slide and we produce new spots of colloidal solution with gold nanoparticles at specific spatial confining.

Micro-heater system
The micro-heater is made of a pointed wire of tungsten (300 µm thick). The thermal stimulus was controlled by means of a power dissipation driven by a traditional voltage generator that is modulated by a conventional 5 V transistor-transistor logic (TTL) signal.

Synthesis of gold nanoparticles
The synthesis of the gold nanoparticles (AuNPs) was carried out in aqueous solution by using the seed-mediated growth method [35] . Briefly, a seeds (~3nm) solution was prepared at room temperature by dissolving a certain amount of HAuCl4 and TC in water to get a final

UV-VIS spectroscopy
The optical properties of AuNPs and their molar concentration were measured by an UV spectrophotometer equipped with single monochromator (V−570 from Jasco, Easton, USA).
Absorption spectra of the AuNPs colloids were collected using a 1.00 cm quartz cell with a scan speed of 400 nm/min in the wavelength range from 300 nm to 800 nm. For the quantitative analysis of the gold amount in the AuNPs solution a set of six standards were obtained by diluting the as-prepared gold nanoparticles solution batch. Complete reduction from Au(III) to Au(0) was assumed.

Raman Spectroscopy
The AuNPs SERS performances in solution were evaluated by Raman spectroscopy using pMA as molecular probe. The Raman and SERS spectra were collected by a confocal Raman Jena, Germany). The EF of the colloids was calculated according to our previous work [35].
The performances of the SERS planar sensor were evaluated by running Raman imaging measurements in the mapping mode: the SERS substrate was placed on a piezo-electrically driven microscope-stage with a x,y resolution of 10 ± 0.5 nm and a z resolution of 15 ± 1 nm.
The stage was scanned at a constant speed in the x-y plane with a 2.0 m step size.

Colloids characterization
The SIA was performed on a selection of TEM micrographs, which provided a total nanoparticles population of ≅ 300 units. Figure 2 shows a representative TEM micrograph of the AuNPs and the results of the SIA. According to our previous work [35] the shape and shape distribution of the AuNPs were evaluated by means of the roundness parameter R obtained from SIA, which is defined as: where, A is the particle area and dmax is the major axis. It was found that 98% of the population is characterized by a spherical shape (R ≥ 0.8); it exhibits an average diameter of 16 ± 3 nm and a unimodal, Gaussian-like size distribution (cf Figure 2B). Only 2% of the total nanoparticle population was characterized by a rod-like shape (R ≤ 0.6). In the whole set of TEM images, no particle aggregation was apparent. The AuNPs colloids exhibit a strong plasmon resonance (SPR) band around 526 nm with a narrow shape, which, according to the SIA results, is indicative of a unimodal distribution of particle sizes (see Figure S1, Supplementary Material). The Au molar concentration in the colloidal solution was evaluated by UV-VIS spectroscopy. The UV-VIS calibration curve was performed by collecting the spectra of six AuNPs standards prepared as explained previously.
An accurate quantitative analysis of the gold amount in the AuNPs batch was possible since the absorbance (at 526 nm) vs concentration plot displays a Beer-Lambert behaviour (i.e., it is linear through the origin, correlation coefficient R 2 = 0.999, cf. Figure S2, Supplementary Material).
After the UV calibration the AuNPs solution was centrifuged and re-suspended in water and an UV absorption spectrum was collected, from which the Au molar concentration was estimated to be 2.6 mM. According to our previous work [35] by coupling the TEM geometrical parameters (size and shape) and the UV-VIS results ([Au]) the concentration of the colloid solution was estimated to be 21 nM. This AuNPs batch was used for the SERS sensor fabrication.
The SERS performances of the synthesized AuNPs in terms of absolute enhancement factor (EF) were evaluated by comparing the spontaneous and the SERS spectra of pMA. In particular, where ISERS is the integrated area of a specific SERS signal (at 1583 cm -1 ) and IREF is the integrated area of the corresponding Raman signal (at 1598 cm -1 ), both normalized for the exposure time. Analogously, NSERS and NREF represent the number of molecules contributing to the SERS and the Raman signal, respectively, and were evaluated according to the procedure described in our previous work [35] and detailed in the Supplementary Material (paragraph 2.1). The estimated EF value of the present colloid is 1.7 × 10 5 , a figure suitable for high-sensitive analytical applications of SERS spectroscopy.

SERS response of the plasmonic nanoparticle assembly
In order to test the SERS activity of the nanoparticle assembly and to verify the signal distribution over the sensing area, we performed a Raman imaging experiment. A (100 × 100) µm area was mapped with a spatial resolution of 2.0 µm; the resulting image, reconstructed by considering the intensity of the pMA peak at 1583 cm -1 , is compared to the visible image of the same region (Figures 3a -3b). The spatially confined assembly is evident in both images as a perfectly round disk with a diameter of 76 µm. In the visible micrograph (Figure 3a) a dark boundary highlights the region where the nanoparticles tend to accumulate (coffee-ring effect).
The Raman image shows a clear correlation between the SERS response and the nanoparticle density. Representative SERS spectra from selected points are shown in Figure 3c. They are highly reproducible, with a flat and consistent baseline denoting the absence of spurious signals and interfering fluorescence. SERS activity is maximized in the boundary region but the pMA spectrum is clearly detected also in areas outside the coffee-ring.