Silver Nanorods Wrapped with Ultrathin Al2O3 Layers Exhibiting Excellent SERS Sensitivity and Outstanding SERS Stability

Silver nanostructures have been considered as promising substrates for surface-enhanced Raman scattering (SERS) with extremely high sensitivity. The applications, however, are hindered by the facts that their morphology can be easily destroyed due to the low melting points (~100 °C) and their surfaces are readily oxidized/sulfured in air, thus losing the SERS activity. It was found that wrapping Ag nanorods with an ultrathin (~1.5 nm) but dense and amorphous Al2O3 layer by low-temperature atomic layer deposition (ALD) could make the nanorods robust in morphology up to 400 °C, and passivate completely their surfaces to stabilize the SERS activity in air, without decreasing much the SERS sensitivity. This simple strategy holds great potentials to generate highly robust and stable SERS substrates for real applications.

Scientific RepoRts | 5:12890 | DOi: 10.1038/srep12890 elevated temperatures 31,32 . For example, coating Ag islands with Al 2 O 3 films could maintain their SERS activity for a period of 34 days 29 ; coating Ag nanorods with TiO 2 thin films by glancing angle deposition (GLAD) technique could sustain their morphology and stabilize the SERS performance at 100 °C (above which coalescence of Ag nanorods was observed) 31 . However, a drawback of this coating approach is the giant decrease in SERS sensitivity, which is caused by the coating layers that separate the target molecules from Ag nanostructures 29,33,34 , and by the possible morphology changes of silver created during coating process. It is therefore highly demanded to find ways to deposit protective layers that completely cover Ag nanostructures at relatively low temperatures to maintain their morphological features, with a precise control of the coating thickness to prohibit sharply decreasing the SERS sensitivity but meanwhile thick enough to make them robust against temperature.
In this study, we employed atomic layer deposition (ALD) technique to prepare ultrathin Al 2 O 3 films that fully wrapping Ag nanorods at a temperature of 50 °C, and investigated the coating influences on the morphological stability of Ag nanorods at elevated temperatures in air, as well as their SERS sensitivity and activity at ambient conditions. It was found that an ultrathin (~1.5 nm) but dense and amorphous Al 2 O 3 layer was thick enough to make Ag nanorods robust in morphology to a temperature of 400 °C, and passivate sufficiently their surfaces to stabilize the SERS activity in air, without largely decreasing the SERS sensitivity. Figure 1a shows a typical SEM micrograph of the as-deposited Ag nanorods, from which one sees clearly that these slanted nanorods are well separated. In order to find an appropriate deposition temperature at which morphology changes of Ag nanorods can be avoided during ALD processing, we heated the as-deposited Ag nanorods in the ALD chamber at temperatues of 50, 80 and 100 °C for 8 minutes, separately, without purging the ALD precursors. Figure 1b-d show the SEM images of Ag nanorods after thermal treatment at the three temperatures. In comparison with the as-deposited nanorods, no noticeable morphology change was observed for those heated at 50 °C. For the samples annealed at 80 and 100 °C, however, partial melting of Ag nanorods was observed, which resulted in adhesion of neighboring nanorods. Since the SERS effect is very morphology-dependent, this may cause sharp differences in their SERS sensitivity 35,36 . Figure S1 compares the average diameters and SERS performance of the above four samples. It is seen that Ag nanorods became larger after heated at 80 and 100 °C and declined in SERS sensitivity, using 5 × 10 −6 M methylene blue (MB) as the probe molecule, while the nanorods heated at 50 °C were of a similar SERS sensitivity as the as-deposited ones. We thus selected 50 °C to prepare Al 2 O 3 coatings by ALD approach. Al 2 O 3 thin films of various thickness were deposited on Ag nanorods at 50 °C by changing the ALD cycle numbers to 1, 2, 3 and 5, respectively. Figure 2a shows a typical top-view SEM image of the Ag nanorods after coating by 5 ALD cycles; inset is a corresponding side-view SEM image. One sees that the nanorods are ~30 nm in diameter, ~280 nm in length, and are well-separated, without apparent variation in morphology compared with the as-deposited ones (see Fig. 1a). Figure 2b shows the HRTEM images of individual Ag nanorods coated with Al 2 O 3 layers by 1, 2, 3 and 5 ALD cycles. It is seen that the coating layers are amorphous in structure and of different thickness (from < 1 nm to ~4 nm), fully wrapping the nanorods. A linear relationship between the thickness of Al 2 O 3 layers and ALD cycle numbers was observed and plotted in Figure S2, from which a deposition rate of ~0.7 nm/cycle was derived. This suggests a precise control of the layer thickness to sub-nanometer scale by this approach 33,37,38 . The chemical states of the Ag nanorods and Al 2 O 3 shells were analyzed by XPS. Figure 2c shows a typical Al 2p XPS spectrum (calibrated with reference to the C1s peak at 284.8 eV) of the coating layer deposited by 5 ALD cycles. The peak is located at ~74.5 eV, indicating the formation of Al-O bond in Al 2 O 3 39,40 .

Results and Discussion
In addition, the Ag 3d 5/2 and Ag 3d 3/2 doublet peaks of this coated sample and the uncoated one are both centered at ~367.8 and ~373.8 eV, respectively, see Fig. 2d. This is good in agreement with those of elemental Ag 41,42 , and indicates that there was no destruction of Ag by ALD processing 30,37 and that the coating layer by 5 cycles was very thin.
The influence of Al 2 O 3 layers on the morphological stability of Ag nanorods was investigated by thermally annealing these samples in air at temperatures of 200, 300 and 400 °C for 30 minutes. It was found that the uncoated Ag nanorods melted completely at these temperatures, evolving from irregular  Figure 3b shows the SEM images of Ag nanorods coated by 1 and 2 ALD cycles, after annealing at 300 and 400 °C, separately. The nanorods coated by 1 ALD cycle (the Al 2 O 3 layer was < 1 nm thick) were robust in morphology at 200 °C (not shown), but melted partly at 300 and 400 °C. For the nanorods coated by 2 ALD cycles (the Al 2 O 3 layer was ~1.5 nm thick), however, no obvious morphological change was observed after being heated at 300 and 400 °C, in comparison with the as-deposited ones. Similar results were also obtained for Ag nanorods coated by 3 and 5 ALD cycles, see Figure S3. From literature, it is known that any morphological variation of Ag nanorods may cause sharp differences in their optical reflectance and SERS sensitivity [43][44][45] . Figure 3c,d and S4 show respectively the reflectance spectra of uncoated Ag nanorods and coated ones by 1, 2, 3 and 5 ALD cycles, before/after thermal treatment. Due to the large morphological changes of the uncoated sample after annealing (Fig. 3a), their reflectance spectra were very different from those before annealing. For the nanorods coated by 2 or more ALD cycles, their reflectance spectra remained almost unchanged after annealing, which is in good agreement with their robust morphology shown by Fig. 3b and S3. As for the nanorods coated by only 1 ALD cycle, the reflectance spectrum remained unchanged at 200 °C, and varied slightly but obviously at 300 and 400 °C. This is due to their morphology changes shown by Fig. 3b and indicates that an Al 2 O 3 layer of < 1 nm was too thin to protect the morphology of Ag nanorods at temperatures > 200 °C.
The SERS performance of the above Ag nanorods before/after annealing was evaluated using 5 × 10 −6 M MB as the probing molecule. Figure 4a compares the Raman spectra of MB on uncoated Ag nanorods as SERS substrate, before/after annealing at 200, 300 and 400 °C, respectively. It is seen that the Raman intensity of MB on annealed, uncoated Ag nanorods dropped drastically, owing to the deterioration of their morphology, see Fig. 3a. For nanorods coated by 1 ALD cycle, see Fig. 4b, their SERS sensitivity remained almost unchanged after annealing at 200 °C, and decreased somehow at 300 and 400 °C because of the slight change of their morphology (see Fig. 3b). For the Ag nanorods coated by 2 ALD cycles, see Fig. 4c, their SERS sensitivity remained almost constant after annealing at 200, 300 and 400 °C. These results suggest that an ultrathin (~1.5 nm) Al 2 O 3 layer by 2 ALD cycles was effective to protect Ag nanorods in both morphology stiffness and SERS sensitivity up to a temperature of 400 °C in air. The effect of Al 2 O 3 coating layers on the SERS activity of Ag nanorods at room temperature in air was also investigated as a function of time. As the starting point, the Raman spectra of 5 × 10 −6 M MB on as-prepared, uncoated Ag nanorods, and on as-prepared, coated Ag nanorods by 1, 2, 3 and 5 ALD cycles were measured and compared as references, see Fig. 5a. As expected, all coated nanorods exhibited strong SERS sensitivity, and a gradual decrease in the Raman signal intensity of MB was observed with an increase in the thickness of Al 2 O 3 coating layers 32 . Afterwards, the Raman spectra of MB were measured again in every 10 days using the above samples as SERS substrates (stored in air). Figure 5b,c show Raman spectra of MB on uncoated nanorods and on nanorods coated by 2 ALD cycles, respectively, measured at different times. One sees that the SERS sensitivity of the uncoated substrate dropped rapidly in air, while that of the substrate coated by Al 2 O 3 by 2 ALD cycles remained almost constant. For a clear comparison, we normalized the intensity of the peak of MB at 1622 cm −1 obtained on various SERS substrates to that obtained on the as-prepared, uncoated Ag nanorods, and plotted in Fig. 5d as a function of time. It shows that the SERS sensitivity of uncoated Ag nanorods dropped drastically in air, and was about one order smaller after 50 days, while that of nanorods coated by 2 and more ALD cycles remained at the same level. This indicates that the stability of the SERS sensitivity of Ag nanorods in air was greatly improved by Al 2 O 3 coating. It is also noticed that after coating with an Al 2 O 3 layer by 2 ALD cycles (~1.5 nm thick), the SERS sensitivity of Ag nanorods was still ~50% that of the as-prepared, uncoated ones. For nanorods coated with a thinner layer (< 1 nm thick) by only 1 ALD cycle, their SERS sensitivity was a little higher at the beginning, but decreased ~18% after 50 days in air, i.e., their SERS stability was not as good as that of Ag nanorods coated by 2 ALD cycles or more. These suggest that a ~1.5 nm Al 2 O 3 layer be optimal to combine both excellent SERS sensitivity and outstanding SERS stability of Ag nanorods in air.
The reason why an Al 2 O 3 coating layer deposited by 1 ALD cycle (< 1 nm thick) was not capable to effectively protect Ag nanorods is intriguing. It is known that pyridine has a small size and a relatively large Raman cross-section and can be adsorbed on the surface of Ag nanostructures intensively. Therefore, it can be used as a mark to check out whether the Al 2 O 3 coating layers were compact enough to protect Ag nanorods. If there are any pinholes in the Al 2 O 3 layer, pyridine could penetrate through pinholes and be adsorbed on the surface of Ag nanorods, thus exhibiting Raman siganls [46][47][48] . We dipped the as-prepared Ag nanorods (uncoated, and coated by 1, 2, 3 and 5 ALD cycles) into a 1 × 10 −2 M pyridine solution, and measured the Raman spectra. Figure 6 shows Raman spectra of pyridine on different substrates. It is seen that Raman spectra of pyridine showed up only on uncoated Ag nanorods and on nanorods coated by 1 ALD cycle. On those coated by 2 and more ALD cycles, no Raman signal of pyridine was observed. These suggest that an Al 2 O 3 layer of < 1 nm thick (by 1 ALD cycle deposition) was too thin to avoid pinholes, and was not dense enough to protect effectively Ag nanorods, and confirm again that an Al 2 O 3 layer of ~1.5 nm thick was an optimal choice by this approach.
In conclusion, we developed a low-temperature ALD approach to completely cover Ag nanorods with thin but dense and amorphous Al 2 O 3 layers, and found that an Al 2 O 3 layer of ~1.5 nm thick was effective to preserve the morphology of Ag nanorods to a temperature of 400 °C, and stabilize their SERS sensitivity in air for at least 50 days. The present study provides a simple way to fabricate Ag nanostructures with both excellent SERS sensitivity and outstanding SERS stability in air, which could be used in real SERS-based sensors.

Methods
Fabrication of Ag nanorod substrates. Slanted Ag nanorods were prepared on Si (001) substrates by oblique angle deposition (OAD) technique in an electron-beam system (GLAD, Thermionics Inc.) with a background vacuum level on the order of 10 -6 Pa. During deposition, the incident angle of the vapor flux was set at ~86° off the surface normal of substrates, with a deposition rate of 0.75 nm/s. The deposition stopped at a thickness of 500 nm read by a quartz crystal microbalance (QCM). The detailed deposition procedure can be found elsewhere [49][50][51][52] . Characterization of Al 2 O 3 -coated Ag nanorods. The morphology, structure and chemical states of Ag nanorods and the Al 2 O 3 layers were characterized by scanning electron microscope (SEM, JEOL-JMS-7001F), high-resolution transmission electron microscope (HRTEM, JEOL-2011) and X-ray photoelectron spectroscopy (XPS, PHI 5300) with Mg Kα as the excitation source, respectively. The reflectance spectra were measured by a R1 angle-resolved spectroscopy system (Idea Optics Co.) in the 200-1000 nm range, using a mirror-like Al film as the reference. SERS detections. The SERS measurements were conducted with an optical fiber micro-Raman system (i-Raman Plus, B&W TEK Inc.), using methylene blue (MB) and pyridine as the probing molecules. Before SERS measurements, the substrates (uncoated and coated Ag nanorods, before/after annealing) were merged into 5 × 10 −6 M MB or 1 × 10 −2 M pyridine aqueous solutions for 30 minutes, and dried naturally in air. The Raman spectra were obtained using a 785 nm laser as the excitation source, with its beam spot focused to ~80 μ m in diameter and using an excitation power of 120 mW. The data collection time of one spectrum was set to 10 s. For every sample, the spectrum was obtained by averaging the spectra obtained from five different areas of the SERS substrate.