All reagents used were of analytical grade. The sputtering target used for the deposition of the base films consisted of zinc oxide (99.99% purity, 0.15" thick and 2" diameter) produced by Superconductor Materials, Inc. The p-type Si(100) and indium tin oxide (ITO; 15 - 25 Ω/in²) substrates were purchased from Sigma Aldrich Chemical Co. (Milwaukee, WI), hexamethylenetetramine (HMT), 99.99%, and zinc nitrate hexahydrate (Zn (NO₃)₂∙6H₂O) were purchased from Sigma-Aldrich and were used to synthesize and deposit ZnO rods onto the above-mentioned sputtered base films. Gold (III) chloride hydrate (99.999% trace metals basis) was obtained from Sigma-Aldrich Chemical Co during this research and used for the preparation Au-coated ZnO NR.

We have previously reported the growth of ZnO rods onto ZnO/ITO base film substrates by wet chemistry following Vassiere’s protocol [16]. This material was utilized to develop an urea biosensor and demonstrated the proposed technique to be successful [21]. In this work we conducted a systematic study of the deposition of ZnO NR, exhibiting different thickness, on ZnO base films deposited over ITO and Si(100) substrates by magnetron sputtering.

ZnO base films were grown using an ATC Orion 8 Linear Magnetron Sputtering instrument (AJA International, Inc., North Scituate, MA) of on-axis configuration, using 150 W of RF power and a deposition time of 10 min. The optimum position on substrate holder of the substrates with respect to the target distance was 8.0 cm. The base film growth took place under an argon atmosphere at 1.0 × 10⁻³ Torr and 120˚C ± 2˚C. The temperature was measured using a thermocouple, which was in contact with the top plate of the substrate holder.

ZnO rods were synthesized from the thermal decomposition of zinc nitrate hexahydrate/hexamethylenetetramine (HMT) in aqueous phase [14,20,21]. The ZnO base/Si substrate and ZnO base/ITO substrates, deposited by RF magnetron sputtering, were immersed in a glass beaker containing Zn (NO₃)₂·6H₂O and HMT at a concentration of 0.01 M. ZnO base films were placed vertically at the bottom of the glass vessels. The reacting solution in contact with the films was refluxed at 95˚C for 1.5 hr. At the end of the contact time, treated films were removed from the solution and thoroughly washed with high purity water (Milli-Q, 18.2 MΩ·cm) to remove any residual amino complex. All ZnO rods were dried at 200˚C for 10 min and then characterized.

A UV light lamp was used to obtain Au-coated ZnO NR in order to detect various analytes. The experimental procedure was carried out using short-wave UV light based on the photoreduction of gold (III) chloride hydrate (HAuCl₄), from a 1 × 10⁻³ M solutions with high purity water (MilliQ 18.2 MΩ·cm). ZnO NR films of 1 cm × 1 cm were swamped in a beaker with 4.0 mL of HAuCl₄ solution at pH 7.2, during 2 hours of reaction time under exposition to UV light. Afterwards, films were rinsed with high purity water and dried at 120˚C for 2 hours to be used as substrates for SERS.

For SERS analysis, ZnO NR was characterized using 4-ABT as SERS probe molecule. An aliquot of 2 μL of a solution 1 × 10⁻⁸ M of the analyte in ethanol was prepared and the pH adjusted to 7.5 and mixed with 100 mL of 0.1 M sodium chloride (NaCl). This mixture was sonicated for a period of 15 min followed by filtration through a 0.5 μm mesh. SERS spectra were recorded at the same conditions for: 4-aminobenzenethiol (4-ABT) and 1,3,5-trinitroperhydro-1,3,5-triazine (RDX). Raman spectra for each mixture was recorded using 785 nm excitation wavelength in a Renishaw Raman Microspectrometer with a 10× magnification and 170 mW head power.

X-Ray diffraction (XRD) measurements were carried out in a Siemens D5000 XRD using the Cu-K α radiation (1.5405 Å). Scanning electron microscopy (SEM) analyses were conducted on a JEOL-JSM 6500 instrument and a Philips/FEI, XL30 s, FEG SEM/Phoenix EDAX. Optical absorption measurements of ZnO rods were carried out with a Beckman DU 800 UV-VIS spectrophotometer. Photoluminescence (PL) spectra were acquired at room temperature (RT) with an HORIBA Jobin/Yvon FluoroLog-3 spectrofluorimeter at an excitation wavelength of 350 nm.

The X-ray diffraction patterns corresponding to the ZnO/ ITO substrates and the ZnO deposited onto this base are shown on Figures 1(a) and (b), respectively. In both samples, the diffraction patterns clearly show the preferential growth along the (002) plane in the wurtzite structure of ZnO. However, some ZnO NR did not grow in the c vertical axis due to the weaker appearance of the (101) and (102) planes. This suggests the formation of wellorganized columnar structures along c-axis direction perpendicular to the substrate. The peak at 35.4˚ corresponds to the ITO substrate.

The broadening of the XRD peaks evidenced the nanocrystalline nature of the ZnO NR grown. The average crystallite size, estimated by the Scherrer’s equation, was about 26 nm. Although both XRD patterns evidenced the preferential growth in c-axis direction (intensities of the (002) peak), the later was shifted toward lower diffraction angles in the pattern with respect to the standard value at 2θ = 34.422˚ (ICDD Powder Diffraction File #36-1451), as shown in Table 1 and in Figure 1(a). This shift can be attributed to tensile stress in the ZnO film deposited by sputtering onto the ITO substrate. It is also apparent that for the (002) plane, there is a slight shift from the standard value (2q = 34.422˚) [23]. The diffraction peaks can be indexed to a Wurtzite hexagonal structure of the bulk ZnO crystal with lattice constants for a = 0.32 nm and c = 0.51 nm from the orientation plane (002).

Table 1 summarizes XRD results and estimated crystallite size for ZnO thin films on the ITO and Si substrates, and for ZnO NR on ITO, Si(100). Note that the strain is tensile between ZnO and ITO and is compressive between ZnO and Si.

Table 1. Lattice parameters of ZnO thin films deposited on different substrates.

Figure 2 shows the UV-VIS optical absorption spectrum for the ZnO rods grown on the ZnO/ITO substrate. The excitonic absorption peak is clearly observed about 364 nm, which exhibits a blue shift compared to bulk ZnO materials (388 nm) [24]. This blue shift observed may be attributed to a confinement of the ZnO rods to nanoscale dimensions [25]. Optical absorption analysis was not possible for the Si substrate due to the lack of transparency for the ZnO NR deposited on ZnO/Si(100) substrate in the UV-VIS spectral range of interest.

The photoluminescence spectra of ZnO NR deposited onto ZnO/ITO base films substrates are shown in Figure 3. The spectrum was recorded at room temperature and under an excitation wavelength of 350 nm. The spectrum clearly displays the dominant narrow near-band-edge (NBE) at 380 nm and also a weaker relative intensity a broad green luminescence (GL) at 550 nm emission band previously reported in ZnO thin films [25,28]. Even though the origin of the GL band has been a matter of discussion [26-28] it is typically associated with deeplevel defects such as oxygen vacancies and interstitial Zn ions, which is largely dependent on the synthetic route used for preparation of the ZnO NR.

Figure 4 shows the diffraction patterns of ZnO/Si(100) base film substrates (4a) and the corresponding ZnO rods deposit (4b). In this case, the base film was produced after a sputtering time of 10 min. The peak broadening observed in the XRD pattern of ZnO base film suggests indicates a smaller crystallite size, with respect to those presented in Section 3.1. The estimated average crystallite size using Scherrer’s formula was 9 nm, which is larger than the 4 nm estimated for the ZnO base film deposited onto ITO substrates. This is mainly due to different types of ZnO stress on these substrates. Also, in this case, the exclusive presence of the (002) diffraction peak evidences the preferential growth of (002) planes in both the ZnO base film and the ZnO deposit. Furthermore, the preferential growth of the (002) plane and the absence of other planes different from those corresponding to the (001) plane in the ZnO deposit confirm the strong out-of-plane orientation for ZnO NR/ZnO/Si. This suggests the development of columnar crystals.

Shown in Figure 5 are PL results obtained for ZnO rods grown on the ZnO base film/Si(100) substrates. The spectra were recorded at an excitation wavelength of 350 nm, which lies within the absorption profile of the nanostructures. A weak shoulder near 381 nm was observed and was attributed to donor acceptor pair levels in ZnO base on p-type Si(100) substrate [28].

A small blue shift was observed on ZnO NR/ZnO/Si with respect to ZnO NR/ZnO/ITO corroborates XRD results. Both Figures 3 and 5 clearly display the dominant narrow NBE and also the broad GL emission bands with lower relative intensity characteristic of ZnO rods [25,28], at 380 nm and 550 nm, respectively. As observed, the qualitative features of the ZnO NR spectra are similar in both surfaces, regardless of the nature of the

substrate, mainly due to the use of ZnO base thin films on those substrates. Moreover, the relative PL intensity ratios of the UV NBE emission to the GL band (I_UV/I_vis) are 0.3 and 0.4 for the ZnO base film/ITO and ZnO base film/Si(100) substrates, respecttively. Thus, the similar I_UV/I_vis ratios observed herein are indicative of the reliability of the preparation procedure for yielding ZnO NR in a consistent manner for both the ITO and Si(100) substrates. Regarding the ZnO/Si(100) and ITO base films, significant emission could not be recorded, due to the lack of nanostructure on base thin films.

In order to attain additional insights on the morphological features of our samples, prepared base films and ZnO deposits were analyzed by field-emission scanning electron microscopy.

FE-SEM images shown in Figure 6 correspond to the ZnO/Si(100) base films grown by RF magnetron sputtering. The films appear highly compact, which suggest a good quality deposition. The grains look equiaxial. These grains are considered as seeds for the growth of the corresponding rods.

The FE-SEM images shown in Figure 7 correspond to the ZnO/ITO base film. As with the Si(100) substrate, the films also seem uniform and compact again suggesting

good sputtering deposition quality. When comparing Figures 6 and 7, the morphological features in both types of ZnO base films are evident. While ZnO grains were more or less equiaxial when ZnO was deposited on Si(100), they exhibited elongated shape when ITO was used as substrate. The crystal growth on the Si(100) substrate is a function of a particular orientation of the Si atoms on a specific plane. The size of the structures varies by orders of magnitude, from nonporous silicon to mesoporous Si, with crystallites in the nanometer scale. The diverse morphological features of porous Si can be affected by pore orientation, filling of macropores and depth variation of porous Si [29]. However, ITO substrates do not have an orientation-specific plane and they typically have low roughness upon manufacture. These structural differences affect the morphology of ZnO base film deposited by RF magnetron sputter on Si(100) and ITO substrates.

Figure 8 shows the FE-SEM images of the ZnO deposited on ZnO/Si(100), Figures 8(a) and (b) and ZnO/ ITO Figures 8(c) and (d) base films substrates. The rodshaped particles are clearly noted. Rods deposited onto ZnO/Si(100) base films averaged 60 nm in thickness with lengths above 3 μm. ZnO rods grown onto ZnO/ITO base films were also nanometric in thickness (~70 nm) although shorter (up to 500 nm in length) compared to the rods grown on ZnO/Si(100) base films. These small differences in morphology of the ZnO NR can be attributed to the observed differences on the morphology (grain size and shape) of the base films [30].

The appropriate selection of the synthesis conditions, such as temperature and concentration of the reactants in aqueous phase, allowed a certain degree of control on the crystal growth rate and the final morphology of the NR obtained. HMT plays a key role in the growth of rodshaped ZnO nanostructures contributing to form a 7.2 pH buffer. HMT would thermally decomposed to release OH⁻ ions as required for the formation of intermediate Zn(OH)₂ [12] and subsequent dehydration to form the corresponding oxide. Regarding particle shape, the preferential adsorption of amine groups (from HMT) on specific crystallographic planes of ZnO has been reported, promoting the preferential growth along the “c” axis of the hexagonal Wurtzite unit cell, which explains the rod-shaped morphology observed by FE-SEM analyses [31].

Compared to Vayssieres’ method, our approach favors formation of NR of thinner diameter (~60 nm) and shorter length (<3 mm) [16]. It has been observed that NR covered with a thin layer of gold of these dimensions exhibit SERS performance with the possibility of molecule-level detection of different analytes [6].

Diffraction patterns for ZnO NR/ZnO base film/ITO are

shown in Figure 9(a). The three orientation planes (100) for ZnO and (111) and (200) for Au structures are observed. The (002) plane is absent, suggesting deformation of columnar structures along the c-axis in a direction perpendicular to the substrate for ZnO NR. Usually the (002) plane corresponds to a maximum intensity of ZnO zincite Wurtzite structure.

The broadening of the XRD peak evidenced the nanocrystalline nature of the ZnO deposit. The average crystallite size, estimated by the Scherrer’s equation, was around 4.0 nm. Au-coated on ZnO NR/ZnO base film/ Si(100) shows a rise with respect to the (111) plane and signals also appear for the (200) and (220) planes characteristic of the fcc Au phase [32,33], as can be observed on Figure 9(b). Reduction of the (002) planes with respect to (111), confirmed the deposition of Au on ZnO NR. Wamng et al. [34] have studied the electron transfer of the ZnO/Au system under UV light and concluded that the electrons transfer from Au to ZnO to form a uniform Fermi energy level during the formation of the Au-ZnO junction, as gold has a higher Fermi energy level than ZnO. Transferred electrons accumulate on the equilibrated Fermi level near the bottom of the conduction band of ZnO. Thus, more electrons can drop from the conducting band of ZnO to the valence band and recombine with holes, so more photons are radiated.

Figure 10 shows the Raman spectrum for Au-coated ZnO NR/ZnO base film/Si(100) at 785 nm. The NR was active for in this excitation line. The band at 521 cm⁻¹ is attributed to the silicon surface signal. A broad peak appears from 1500 up to 3000 cm⁻¹, from photoluminescence due to the gold coating.

4-ABT was selected as SERS probe molecule because

it absorbs strongly on gold from a gold-sulfate bond [34] activity of the ZnO NR. Figure 11 shows SERS of 4-ABT on Au coating ZnO NR/ZnO base film/Si obtained by both methods, ion beam sputtering and UV light lamp irradiation. A peak at 390 cm⁻¹ confirms that 4-ABT molecules are bound via a gold-sulfate bond. The Au-S mode increased in intensity with concentration from 1 × 10⁻⁸ M up to 1 × 10⁻⁴ M of 4-ABT. This last result indicates that 4-ABT was oriented perpendicular to the Au-coated ZnO NR surface [33]. The enhancement of intensities of the peaks correspond to: C-S stretching (1075), C-H stretching (1171), C-H bend and C-C bending (1374) and the C-C stretching 1588 cm⁻¹ [34]. These are mainly due to the field enhancement effect induced by surface plasmon polaritons of Au-coated ZnO NR [1].

Strong SERS of Raman spectra was observed for 4-ABT. However the experiments were realized with addition of Cl⁻ ions to increase the ionic strength effect. A limit of detection (LOD) of 1 ×10⁻⁸ M for ABT was

achieved corresponding to a minimum of 5.4 × 105 molecules/cm² detected under the experimental conditions at excitation wavelength of 785 nm with a sensitiveity of the ZnO NR in the range of 1.1 ×10⁻¹⁶ g under laser spot, with a SEF of 7.5 × 108. The enhancement signal was induced by the chemical effect, due to an electronic resonance transfer between absorbed molecules and the metal surface [35]. The contribution of chemical enhancement is in the order of 10¹ to 10². Therefore the electromagnetic factor can be attributed to 10⁶ up to 10¹⁴. Both effects contribute to high SERS enhancements factors for Au-ZnO NR [36].

Figure 12 shows SERS spectra at different concentrations of RDX on Au-coated ZnO NR/ZnO base film/Si (100). The development of this method allowed for lowering of the detection limit of RDX in thin films. A similar procedure was used in both samples dissolved in ethanol prepared at an adjusted pH = 7.5 and mixed with 100 mL of 0.1 M sodium chloride (NaCl). Acquisitions were obtained at 3 accumulation, 10 s integration time using a laser power of 785.0 nm, 98 mW measured at head.

The 874 cm⁻¹ peak diminished in I_SERS to a concentration of 1 × 10⁻⁴ in comparison with the concentration of 1 × 10⁻⁶ M of RDX components. Low detection limits were found for this analyte (9 × 10⁻¹⁹ g under spot of laser).

Raman-bands assignments of RDX are as follows: the vibration mode at 930 cm⁻¹ is assigned to ring stretching and NO deformation signal intensity in SERS. The band at 874 cm⁻¹ is attributed to the symmetric ring-breathing mode; the band at 1258 cm⁻¹ is due to CH₂ scissoring and N-N stretch vibration; [37] the band at 1312 cm⁻¹ results from CH₂ wagging; the band at 1370 cm⁻¹ is υ NO₂ symmetric stretching vibration and β CH₂ scissoring. Finally the band at 1560 cm⁻¹ is attributed to the υ NO₂ asymmetric stretch in nitramines [37].