High-temperature stability of silver nanoparticles geometrically confined in the nanoscale pore channels of anodized aluminum oxide for SERS in harsh environments

Abstract

We report the ability of nanoscale pore channels of anodized aluminum oxide (AAO) to endow entrapped silver nanoparticles (Ag NPs) within with structural and oxidation stability for potential surface-enhanced Raman scattering (SERS) at elevated temperatures. AAO was prepared via two-step anodization of high purity aluminum foil in phosphoric acid. Ag NPs of controlled size and coverage were obtained via in situ seeded growth from aqueous AgNO₃ solution inside the AAO pore channels. The structural and chemical characteristics and the SERS activity of the Ag NPs before and after environmental exposure in air at up to 600 °C for as long as 5 days were evaluated using scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), and Raman spectroscopy. We show that the Ag NPs entrapped in the AAO pore channels exhibit enhanced structural and oxidation stability and thus retain significant SERS activity upon high-temperature treatment, indicating the intricate role of geometric confinement in retarding Ostwald ripening, evaporation loss, as well as oxidation of Ag NPs.

---

Introduction

Surface-enhanced Raman spectroscopy (SERS) enabled by noble metal nanostructures such as Ag and Au has long been exploited for chemical sensing and detection at trace concentrations.^1,2 Thermal instability of these nanostructures and hence degradation of SERS activity^3–7 poses a major challenge for their application in high-temperature environments such as those encountered in fuel gas streams of advanced fuel cells and coal gasifiers. Numerous efforts have been made to overcome this challenge. For example, the Van Duyne group demonstrated that a triangular Ag nanostructure capped with refractory Al₂O₃ thin film via atomic layer deposition retained its shape and SERS activity after thermal annealing at 500 °C in an inert environment.⁸ Using a similar deposition method, Dai et al. showed improved structural stability of Al₂O₃-coated Ag nanowires at 400 °C and a SERS enhancement factor (EF) of 5.5 × 10⁴.^9,10

The Al₂O₃ thin film strategy offers an effective remedy to the thermal instability of Ag nanostructures at elevated temperatures. This approach is at the expense of SERS enhancement. For instance, the presence of a protective coating prevents the adsorption of analyte molecules directly on the Ag surface, making the chemical enhancement mechanism inoperative. In addition, unless the coating is extremely thin, on the order of a few nanometers, the hot-spot effect between adjacent nanostructures, the primary source of electromagnetic field enhancement for ultra-sensitivity, will be suppressed. It is well established that SERS sensitivity is optimized when the chemical enhancement and electromagnetic enhancement in hot spots are in effect.^11,12 Any development of a thermally stable SERS platform without surface passivation or encapsulation of Ag nanostructures will significantly advance the application of SERS for harsh environments.

Here, we explore the potential of the nanopore channels of anodized aluminum oxide (AAO) to impart thermal stability on Ag NPs. The superior chemical and structural stability of AAO, compared to mesoporous silica, makes it a natural host of Ag NPs for SERS in harsh environments. The use of AAO as a SERS substrate also takes advantage of the fact that its vertically-aligned pore channel structure exhibits optical waveguiding characteristics along the pore walls of AAO for stronger evanescent field interaction with Ag NPs that could lead to further enhanced Raman sensitivity, as reported by Tsukruk's group in their room temperature measurements.^13–15 Our work entails anodization of high-purity aluminum foil to fabricate AAO with well-organized nanopore channels, in situ growth of Ag NPs inside the channels, and thermal treatment of the AAO with entrapped Ag NPs at up to 600 °C for as long as five days in conjunction with comprehensive studies of the structural and chemical characteristics of the AAO support and Ag NPs within, and their use as a SERS-active platform for Raman measurements of rhodamine 6G (R6G) solution (∼10⁻⁶ M) before and after the thermal treatments. We show that the size and distribution of Ag NPs inside the nanopore channels are well-preserved and the Ag NPs are less susceptible to oxidation under the conditions investigated, compared to Ag NPs on the surface of the AAO substrate. More importantly, Ag NPs within AAO exhibit strong SERS activity after prolonged exposure at high temperatures.

Experimental

Materials

All chemical reagents were purchased from commercial vendors and used without further purification: tin chloride dehydrate (ACS reagent, 98%, Sigma-Aldrich), hydrochloric acid (ACS reagent grade, 36.5–38%, PHARMCO-AAPER), silver nitrate (ultrapure grade, Acros), ascorbic acid (ACS reagent, ≥99%, Sigma-Aldrich), rhodamine 6G (R6G; Sigma-Aldrich), pyridine (ACS reagent, ≥99.0%, Sigma-Aldrich), high purity aluminum foil (0.1 mm thick, 99.99%, Alfa Aesar), rectangular aluminum bar (alloy 6063, rounded edge, 1/8′′ thick, McMaster-Carr), ethanol (anhydrous, 99.8%, Sigma-Aldrich), perchloric acid (60–62%, Acros), phosphoric acid (85.3%, Acros), chromic oxide (CrO₃; ACS reagent, ≥99%, Sigma-Aldrich). Water used for experiments was filtered with Barnstead ion-exchange columns and then further purified by passage through Millipore (Milli-Q) columns.

Preparation of AAO substrate

AAO was prepared via two-step anodization of high purity aluminum foil. Briefly, aluminum foil was first cleaned by sonication in acetone and then in Milli-Q water for 10 min. It was electropolished in a mixture of perchloric acid and ethanol (with volume ratio of 1 : 4) under 10 V for 70 s at room temperature. Anodization was subsequently conducted in 0.3 M phosphoric acid under 60 V for 2 h at 0 °C in an ice-water bath. Removal of the first anodization layer was employed by soaking in a mixture of 0.4 M phosphoric acid and 0.2 M chromic acid at 60 °C. An identical anodization process was repeatedly carried out. Finally, the pores were widened by chemical etching in 0.3 M phosphoric acid at room temperature. During anodization, the electrolyte solution was vigorously stirred. The applied voltage and current were monitored by a Keithely 2420 source in conjunction with the LabVIEW program. The resultant AAO substrate has an average pore size of 160 ± 20 nm and a thickness of 5 μm.

In situ growth of Ag NPs inside AAO

The AAO was immersed in an aqueous mixture of SnCl₂ (0.02 M) and HCl (0.02 M) for 2 min resulting in the deposition of Sn²⁺ (–O–Sn–O– bond was formed through tin chloride sensitization on the alumina surface) on the pore channels.¹⁶ The AAO was then soaked in 0.02 M aqueous AgNO₃ solution for 2 min to induce Ag seeds on the pore channels. The Ag seed deposition step was repeated up to five times to allow control of particle coverage density. The AAO with Ag seeds was afterwards immersed in an aqueous mixture of 10 mM AgNO₃ and 100 mM ascorbic acid for 2 h to allow growth of Ag NPs from the seeds. Following each step mentioned above, the AAO with entrapped Ag NPs was rinsed in Milli-Q water and subsequently in acetone and dried in oven at 50 °C for 1 min.

Characterization of Ag NPs in AAO

The size and distribution of Ag NPs inside AAO before and after annealing were examined with a scanning electron microscope (SEM, Auriga Zeiss). Elemental analysis was carried out using energy-dispersive X-ray analysis (EDX). Statistical analysis of size and coverage density of Ag NPs inside AAO was done by ImageJ software. Mean free path of silver vapor atom was calculated using equation , where l is mean free path, K_B is Boltzmann constant, T is temperature, σ is diameter of silver vapor atom, and p is the ambient pressure. The chemical state of the Ag NPs in AAO after heat treatment was investigated using X-ray photoelectron spectroscopy (XPS). Sputtering in conjunction with XPS was conducted using a 2 keV Ar⁺ ion beam for 20 min to achieve the removal of a ∼50 nm thick top layer based on calibration for sputtering of Al₂O₃. The XPS measurements were carried out with a PHI 5600ci instrument using monochromatic Al Kα X-rays. The pass energy of the analyzer was 23.5 eV and the scan step size was 0.1 eV. Binding energies were referenced to the Al 2p peak of the substrate, which was assigned a binding energy of 74.4 eV. A PHI model 04-090 charge neutralizer was employed to minimize the effects of sample charging. The neutralizer was operated with an emission current of 19.9 mA and electron energy was set at 9%. The settings were kept constant throughout the study for consistency.

Heat treatments of Ag NPs in AAO

Annealing treatments were done by placing the samples inside an open-ended quartz tube (to avoid contaminants) in the Mellen Microtherm MT furnace in ambient air. The samples were transferred to a vial filled with Ar immediately after the annealing treatments for XPS analysis.

Raman measurements

Raman measurements were performed using a custom-built Raman spectrometer¹⁷ at an excitation wavelength of 532 nm with power intensity of 10 mW. The AAO samples were soaked in an aqueous solution of 10⁻⁶ M R6G for SERS and 10⁻³ M R6G for normal Raman measurements, respectively, for 5 min and then rinsed with water and dried by flowing nitrogen. Vapor phase Raman/SERS measurements were done by exposing the AAO samples to a pyridine solution. Raman spectra were collected with an exposure of 20 s and at least three measurements at different spots were made and the results averaged. Raman data was processed using Origin 8.5 software. The enhancement factor (EF) was calculated using EF = (I_SERS/C_SERS)/(I_NR/C_NR), where I_SERS and I_NR are the integrated intensities of a characteristic band from SERS and from normal Raman, and C_SERS and C_NR are concentrations of analytes used in SERS and normal Raman experiments, respectively. SERS and Raman spectra used for EF calculation (see Fig. S11†) as well as the corresponding calculation process are provided in the ESI.†

Results and discussion

Colloidal Ag NPs are often prepared by citrate reduction of silver nitrate solutions. Uniform immobilization of these post-synthesized Ag NPs in the nanoscopic AAO pore channels is challenging. We utilized seed-assisted in situ growth of Ag NPs within the pore channels instead.¹⁴ Compared to the method based upon capillary force, which frequently induce the deposition of larger nanoparticles predominantly in the AAO pore opening and with reduced concentration along the AAO pore channels, this method provides a uniform density of nanoparticles along the cylindrical nanopores. In our method, Ag seeds were first formed on the channel surface via electroless deposition.¹⁸ As shown in Fig. 1a, this process resulted in dense and uniform distribution of Ag seeds about 9.1 ± 1.7 nm in diameter. These seeds grew to large Ag NPs shown in Fig. 1b–(d) in aqueous solution of AgNO₃ and ascorbic acid of various AgNO₃ concentrations in 2 hours. The sizes of Ag NPs are 10.3 ± 3.1 nm, 13.3 ± 4.7 nm, and 15.6 ± 5.6 nm for 6.7 mM, 7.3 mM, and 8 mM of AgNO₃, respectively. The trend of larger Ag NPs with more concentrated AgNO₃ solutions is consistent with published reports.¹⁹ AAO substrates with Ag seeds as well as subsequently grown Ag NPs were employed for SERS measurements of 10⁻⁶ M R6G using a custom-built Raman spectrometer.¹⁷ The corresponding Raman spectra are illustrated in Fig. 2. No characteristic Raman bands of R6G were detected using AAO containing Ag seeds only, except a broad background from the AAO substrate. Ag NPs developed from the seeds exhibited increasing SERS activity with their size. The SERS enhancement factor (EF) (calculation method described in the Experimental section) is about 2.1 × 10⁵, 4.3 × 10⁵, 5.0 × 10⁵ for Ag NPs with 10.3 ± 3.1, 13.3 ± 4.7, 15.6 ± 5.6 nm in diameter, respectively. We note that the EF generally increases with nanoparticle size up to 100 nm in diameter.²⁰ A higher EF is thus expected with larger Ag NPs in AAO.

Fig. 1 Cross-sectional SEM images of (a) electroless-deposited Ag seeds and in situ grown Ag NPs in AgNO₃–ascorbic acid solutions of (b) 6.7 mM, (c) 7.3 mM, and (d) 8 mM AgNO₃.

Fig. 2 Raman spectra of 10⁻⁶ M R6G using AAO entrapped with (a) Ag seeds and in situ grown Ag NPs in AgNO₃–ascorbic acid solutions (b) 6.7 mM, (c) 7.3 mM, (d) 8 mM AgNO₃. Raman excitation wavelength is 532 nm, laser power 10 mW, and acquisition time 20 s.

Given the same particle size, more densely populated Ag NPs provide increased SERS-active sites and hence the hot spot effect for greater SERS enhancement. In this study, the coverage density of Ag NPs on the pore channels of AAO can be controlled by repeated Ag seed deposition. As illustrated in cross-sectional SEM images in Fig. 3, Ag NP coverage densities of <1/μm², 20/μm⁻², and 80/μm⁻² in the pore channels were obtained upon in situ growth from seeded AAO following one, three, and five repeated seeding steps, respectively. The corresponding spacings between neighboring particles were estimated at 2267 ± 643 nm, 110 ± 103 nm, and 30 ± 21 nm, respectively. The nanoparticles were discretely and evenly distributed along the entire pore channel length of the AAO structure, resulting in a 3D SERS-active platform structure with high detection sensitivity due to its high specific surface area compared with Ag NPs immobilized on the surface of a planar substrate.¹⁵

Fig. 3 Seed coverage density as a function of number of repeated seeding deposition and the corresponding cross-sectional SEM images of in situ grown Ag NPs within AAO pore channels.

We examined the stability of Ag NPs inside AAO before and after annealing at 500 °C and 600 °C from 3 hours to 5 days. The Ag NP size and coverage density were 15 ± 4 nm and 1311 NP/μm², respectively, uniformly and discretely distributed on the AAO pore channel surface in the as-prepared sample, as shown in Fig. 4. After annealing at 500 °C for 3 h, Ag NPs maintained their general morphological features except for some coarsening between adjacent particles, resulting in an increased mean size of Ag NPs at 16 ± 4 nm and lower coverage density of 782 NP/μm². Extending the heat treatment time to 5 days, both the size and coverage density of Ag NPs were reduced to 14 ± 3 nm and 478 NP/μm², respectively. The extended heat treatment led to about 50% total volume or mass reduction of Ag NPs, a strong indication of evaporation depletion resulting from the long exposure at high temperature in ambient air (see ESI Fig. S5†).^21–23 Heat treatments of Ag NPs inside AAO at 500 °C for 20 min, 60 min, and 1 day in ambient air (see ESI Fig. S1†) illustrated the general trend of coalescence of Ag NPs in contact and evaporation loss with time. Annealing of Ag NPs in AAO at 600 °C yielded generally similar trends as seen at 500 °C in terms of coalescence of nanoparticles in close proximity (i.e. particle size grows bigger as annealing time increases) but with intensified evaporation depletion. For example, after a 600 °C-3 hour anneal, the size and coverage density of Ag NPs became 16 ± 4 nm and 516 NP/μm², respectively. After a 600 °C-5 day anneal, the size and coverage density of Ag NPs changed to 23 ± 5 nm and 116 NP/μm², respectively. Heat treatments of Ag NPs inside AAO at 600 °C for 20 min, 60 min, and 1 day in ambient air (see ESI Fig. S3†) yielded similar but intensified results when compared to the experiments at 500 °C. Higher mass loss of the total amount of Ag NPs annealed at 600 °C than that at 500 °C for the same annealing duration was observed (see ESI Fig. S5†), further confirming intensified evaporation with increased temperature.

Fig. 4 (Left column) Plan-view and (center column) cross-sectional SEM images of Ag NPs in the AAO pore channels (a) before and after heat treatment at (b) 500 °C for 3 h, (c) 500 °C for 5 days, (d) 600 °C for 3 h, and (e) 600 °C for 5 days in ambient air (right column).

While thermal annealing at 500 °C and 600 °C for up to 5 days did bring about some changes in the size and distribution of Ag NPs, the fact that a significant amount of Ag NPs still remained in the AAO pore channels after the various annealing treatment is in sharp contrast to the rapid coalescence and complete loss of Ag NPs immobilized on the surface of planar sapphire (Al₂O₃) substrate after annealing at 500 °C for only a few hours (see ESI Fig. S6†).²⁴ Geometrical confinement of Ag NPs in the pore channels of AAO clearly played an intricate and significant role in stabilizing Ag NPs and preventing their evaporation loss. This phenomenon may be attributed to the fact that Ag vapor can only escape at the top surface through a restricted opening due to the extremely confined volume. Constant collision with the pore wall effectively traps Ag vapor and increases its partial pressure within the porous channels, and when combined with the relatively small volume of the AAO pores can result in a rapid local establishment of solid–vapor equilibrium and retardation of mass loss via vapor phase.²⁴

As an illustrative example, one hypothetical limit of the geometric confinement effect can be envisioned for small diameter AAO pores (i.e. nm range) with pore lengths (i.e. micron range) much larger than the pore diameter and a uniform coverage of Ag nanoparticles (or a Ag thin film) on the AAO pore walls. In the case where the pore volume is sufficiently small and the pore length sufficiently long, one can assume that the equilibrium partial pressure of Ag at the surface of the Ag particles can be approximately established throughout the pore volume and the only source of mass loss in the system is through vapor phase diffusion of Ag out of the pore opening. The rate of evaporative losses of a species in a stagnant gas system or one with laminar flow is typically determined by diffusion across a surface boundary layer, and can be assumed proportional to the equilibrium partial vapor pressure of the species at the surface in question and the area of the exposed surface layer. Taking into account the restricted pore size opening area relative to the large pore surface area contained within, a reduction in rate of mass losses due to evaporation of ∼πr²/2πrl = r/2l can be very roughly estimated as compared to the equivalent distribution of Ag nanoparticles on a dense planar surface where r is the pore radius and l is the pore length. For pore radii of ∼50 nm and pore length of ∼1 micron, such an effect could reduce the evaporation rate of Ag to ∼2.5% of the rate expected for a dense planar surface. As the pore radius to length decreases, the simple estimate would predict a roughly proportional reduction in the rate of mass loss due to evaporation and hence the retardation effect is anticipated to intensify with increased pore channel length and reduced pore channel diameter. An enhanced mass loss of Ag due to evaporation at the higher temperature annealing treatment can be largely explained by the higher equilibrium vapor pressure expected at the surface of the Ag nanoparticles in air (∼3.26 × 10⁻¹³ atm at 500 °C and 4.87 × 10⁻¹¹ atm at 600 °C).²⁵ Rough estimates of the Ag mass loss due to evaporation in air based upon the relative partial vapor pressure of Ag suggest a proportional mass loss rate almost two orders of larger at 600 °C as compared to 500 °C.^26–28 In the case where the total Ag within the pores is insufficient to maintain an equilibrium partial vapor pressure throughout the entire pore volume, the kinetics of diffusion of Ag specie in the vapor phase through the pore structure will begin to play an increasingly important role and is anticipated to result in a reduced rate of mass loss.²⁹ Such diffusional kinetics considerations may likely be the cause of a reduced rate in mass loss observed after high annealing times for the AAO samples investigated here (see ESI Fig. S5†).

Our experimental findings are consistent with the observation of Sun et al.,²¹ who reported enhanced thermal stability of Ag NPs within mesoporous silica matrix at 500 °C for 2 hours though SERS was not a subject of the study. Based on the experimental results presented here and the simple arguments outlined above, the 2D dimension of AAO pore channels (i.e. nano-scale diameter and micro-scale length) can be expected to have a significant influence on the thermal stability of nanoparticles. The full thermodynamic and kinetic description of this effect will be significantly more complicated than the simple concepts described above and must include considerations for the impacts of pore structure and diameter on diffusion of vapor phase into and out of the pores, a more realistic gradient in partial vapor pressure of the evaporating species throughout the pore structure, and any additional impacts of the pore structure on the diffusion boundary layer. A more detailed treatment of the thermodynamic and kinetic effects such as those described only briefly here is anticipated to be the subject of additional future investigations.

Corresponding statistical analysis of the size distribution and coverage density (CD) (in NP/μm²) of Ag NPs inside the AAO pore channels before and after various heat treatments was also conducted shown in Fig. 4. Average size and standard deviation were estimated based on particle-projected planar surface area on the interior wall of the pore channels. The corresponding total volume (through a combination of average size multiplied coverage) of the annealed Ag NPs inside AAO pore channels was calculated at 2.26 × 10⁶ nm³, 1.83 × 10⁶ nm³, 0.74 × 10⁶ nm³, 1.07 × 10⁶ nm³, and 0.71 × 10⁶ nm³, respectively. Mass (or volume) loss of the total amount of annealed Ag NPs as compared to that before annealing was estimated accordingly (see ESI Fig. S5†).

Ag NPs are known to oxidize even under ambient conditions. The process of oxidation can result in precipitous decline in SERS enhancement.^30,31 XPS has proven a powerful tool to study the surface chemistry (and oxidation state) of Ag NPs due to its surface sensitivity as well as sensitivity to the chemical binding nature of the elements of interest.³ Our XPS analysis of AAO with entrapped Ag NPs before and after heat treatments was conducted at primarily two locations – Ag NPs on the very surface of the AAO substrate where they are directly exposed to ambient air and Ag NPs inside the AAO pore channels after sputter removal of approximately 50 nm of AAO. Shown in Fig. 5A are Ag 3d photoelectron spectra of Ag NPs on the surface and in the pore channels of AAO after sputter removal of the AAO surface, before and after annealing at 500 °C and 600 °C for 3 h and 5 days. The corresponding Ag 3d_5/2 photoemission lines are summarized in Table 1. We also explored the Ag MVV Auger spectra (see ESI Fig. S8 and S9†) for the Ag NPs that were subject to the same heat conditions. Curve fitting of all the 3d peaks was carried out to identify the contribution from metal Ag (368.2 eV) and Ag oxide (367.5 eV), respectively (see ESI Fig. S10†).

Fig. 5 (A) Ag 3d photoelectron spectra of Ag NPs on the surface and in the pore channels of AAO 50 nm below, before and after annealing at 500 °C and 600 °C for 3 h and 5 days. (B) Summary of the binding energy of Ag 3d_5/2 peak based on the data in (A). (C) Summary of Ag MVV Auger spectra of Ag NPs on the surface and inside AAO pore channels, before and after heat treatment. Metallic Ag and Ag₂O limits (dotted line) were calculated in ESI (see Fig. S7†).

Table 1 Summary of Ag 3d_5/2 photoemission lines

	Unsputtered, 3d5/2	Sputtered, 3d5/2
As-prepared	367.6	367.6
500 °C, 3 h	368.2	367.6
500 °C, 5 d	368.3	367.6
600 °C, 3 h	368.0	367.6
600 °C, 5 d	367.8	367.9

---

In the “unsputtered” samples, the Ag 3d peak for the untreated one has a binding energy corresponding to Ag oxide. The binding energies for the treated ones increase, suggesting metallic Ag, a rare and well documented exception for the direction of shift involving metals and their corresponding oxidation state, while the Auger peak results suggest increased oxidation state. The differences for the treated ones could be explained by invoking a metal core/oxide shell structure selectively probed as a result of the different kinetic energies of the Auger and 3d photoelectrons. The Auger electrons have relatively low kinetic energy so mostly come from close to the surface of the particle, while the 3d photoelectrons have higher kinetic energy and would be more representative of the core. The Auger spectrum for the untreated “unsputtered” one looks exactly like that of Ag metal (see ESI Fig. S7 and S8†), and also the Auger parameter calculated based on the literature is extremely close to the range for Ag metal.³² However, the 3d peak for the untreated “unsputtered” one was found to be characteristic of Ag₂O; this type of discrepancy has previously been reported in the literature³ and may be attributed to any organic adsorbate, contaminants or intentionally applied that may have the potential to result in the shift in Ag 3d peaks. As shown in Fig. 5C, the annealed Ag NPs inside the pore were less impacted by oxidation than those on the surface, indicating the important role of nanoscale pore channels in mitigating the oxidation effect of Ag NPs geometrically confined within. The oxidation retardation of Ag NPs within the nano-scale AAO pore channels, compared to Ag NPs at the surface of AAO support, can be attributed to restricted diffusion path of oxygen in the annealing environment. Note that although it is well-known that Ag bulk is not expected to oxidize at 500 or 600 °C as bulk Ag oxide will undergo thermal decomposition, Ag NPs have a tendency to oxidize at such high temperatures especially when they are very small due to the inverse size-dependent change in free energy for oxidation.³³ Ag NPs may also further oxidize upon cooling in the ambient air environment to temperatures below which the Ag oxide becomes stable, for which the AAO would also be expected to mitigate against this effect. One needs to be mindful of the fact that, while AAO structure offers the benefit of stabilizing Ag NPs from a structural and chemical standpoint due to geometrical confinement, the same confinement effect may potentially lead to compromised sensing response of vapor species. The platform design needs to take full consideration of the complex interplay of the various factors for optimal measurement sensitivity, response time, and stability in harsh environment.

We compared and contrasted the SERS activity of the Ag NPs inside the AAO pore channels before and after thermal annealing using 10⁻⁶ M R6G as a model analyte. Shown in Fig. 6 are the Raman spectra obtained from the as-prepared SERS substrate and those after annealing at 500 °C and 600 °C for 3 h and 5 days. As illustrated, the samples after annealing at 500 °C and 600 °C for 3 hours exhibited SERS enhancement basically the same as the as-prepared sample. This observation suggests that the coalescence of adjacent Ag NPs occurring at this stage is insignificant to have a notable impact on the overall hot spots available for SERS enhancement. Note that necking between two or more nanoparticles does not necessarily result in elimination of hot spots. The sample after 500 °C-5 day annealing retained substantial SERS enhancement but yielded a marked decrease in Raman sensitivity. This decline probably resulted from a combination of reduction of the hot spots due to fusion of coalesced Ag NPs to a single large particle over prolonged time and oxidation of Ag NPs, with the former likely playing a predominant role. This analysis is indirectly supported by the fact that 600 °C-5 day annealing proved to be too harsh, as the sample no longer showed any SERS activity, as a result of intensified fusion of Ag NPs and their evaporation loss. In fact, additional measurements (see ESI Fig. S2 and S4†) indicated that 600 °C-1 day annealing already led to substantial decline in SERS sensitivity. Our investigation has revealed the significant short-term benefit of geometrical confinement in improving the high-temperature stability of Ag NPs and retaining SERS activity. However, this benefit diminishes for prolonged exposure especially at 600 °C. Our current study involved the use of AAO of relatively large pore channel diameters (∼160 ± 20 nm on average) and large porosity. As a future direction, it will be very interesting to see whether the stability region of entrapped Ag NPs (both temperature and time) can be extended further.

Fig. 6 (A) (a) Raman background of as-prepared sample, Raman spectra of 10⁻⁶ M R6G obtained using (b) as-prepared sample and after annealing under the condition of (c) 500 °C-3 h, (d) 500 °C-5 day, (e) 600 °C-3 h, (f) 600 °C-5 day. (B) Summary of Raman intensity for R6G using the band at ∼1376 cm⁻¹ Ag NPs in AAO before and after the annealing treatments.

Our separate experiments as shown in Fig. 7 also show the ability of using AAO with entrapped silver nanoparticles for vapor phase SERS measurement. Peaks at 1008 and 1036 cm⁻¹ are characteristic of pyridine.³⁴ Comparable SERS intensities of pyridine vapor were detected before and after annealing treatment on silver nanoparticles at 500 °C for 5 days, indicating the critical role of AAO pore channels in preserving SERS functionality of entrapped silver nanoparticles for high-temperature gas sensing. Therefore, the abilities of controlling the nanoparticle size/coverage, tailoring the AAO porous structure, and maintaining the SERS functionality using AAO pore channels as host allow this SERS substrate for sensitive and higher-temperature gas sensing such as those encountered in fuel gas streams of advanced fuel cells and coal gasifiers.

Fig. 7 Raman measurements of pyridine vapor (2–3 wt%) using AAO entrapped with silver nanoparticles before and after annealing at 500 °C for 5 days (similar substrate used for experiments shown in Fig. 6). Raman background of silver nanoparticles entrapped inside AAO is also included.

Conclusions

We have demonstrated that Ag NPs can be controllably and uniformly deposited on the surface of nanoscopic pore channels of AAO in situ and via a seed-assisted growth process. The coverage density of Ag NPs can be adjusted by repeating the seeding steps. We have shown that Ag NPs entrapped in the pore channels exhibit significant structural stability and are less susceptible to oxidation than Ag NPs on the surface of a planar substrate at high temperatures. We have revealed that Ag NPs in AAO retain substantial SERS activity after heat treatment at 500 °C for five days and even 600 °C for three hours. Geometric confinement of Ag NPs in the pore channel structure of AAO has clearly played a significant role in endowing them with the enhanced stability. We speculate, based on our results, that the stability region may well be expanded to higher temperatures and longer durations with reduced dimensions of the pore channels and decreased porosity of AAO. Our results suggest that AAO with entrapped Ag NPs throughout its nanoporous structure holds promising potential for SERS in demanding conditions. Areas of applications may include molecular adsorption and desorption during high-temperature catalysis involving precious metals, and sensing and control of advanced energy generation and production systems.

Acknowledgements

This work was funded by National Science Foundation DMR-1506179. This work was supported in part by an appointment to the National Energy Technology Laboratory Research Participation Program, sponsored by the U.S. Department of Energy and administered by the Oak Ridge Institute for Science and Education.

References

1. K. Kneipp, Y. Wang, H. Kneipp, L. Perelman, I. Itzkan, R. Dasari and M. Feld, Phys. Rev. Lett., 1997, 78, 1667–1670.
2. S. Nie and S. R. Emory, Science, 1997, 275, 1102–1106.
3. P. Pinkhasova, H. Chen, M. W. G. M. (Tiny) Verhoeven, S. Sukhishvili and H. Du, RSC Adv., 2013, 3, 17954–17961.
4. Y. Takahashi and T. Tatsuma, Nanoscale, 2010, 2, 1494–1499.
5. a. Ott, S. Ring, G. Yin, W. Calvet, B. Stannowski, Y. Lu, R. Schlatmann and M. Ballauff, Nanotechnology, 2014, 25, 455706.
6. B. Ingham, T. H. Lim, C. J. Dotzler, A. Henning, M. F. Toney and R. D. Tilley, Chem. Mater., 2011, 23, 3312–3317.
7. S. K. Volkman, S. Yin, T. Bakhishev, K. Puntambekar, V. Subramanian and M. F. Toney, Chem. Mater., 2011, 23, 4634–4640.
8. A. V. Whitney, J. W. Elam, P. C. Stair and R. P. Van Duyne, J. Phys. Chem. C, 2007, 111, 16827–16832.
9. E. V. Formo, S. M. Mahurin and S. Dai, ACS Appl. Mater. Interfaces, 2010, 2, 1987–1991.
10. E. Formo, Z. Wu, S. Mahurin and S. Dai, J. Phys. Chem. C, 2011, 9068–9073.
11. M. Moskovits, J. Raman Spectrosc., 2005, 36, 485–496.
12. C. L. Haynes and R. P. Van Duyne, J. Phys. Chem. B, 2001, 105, 5599–5611.
13. H. Ko and V. V. Tsukruk, Small, 2008, 4, 1980–1984.
14. S. Chang, Z. a. Combs, M. K. Gupta, R. Davis and V. V. Tsukruk, ACS Appl. Mater. Interfaces, 2010, 2, 3333–3339.
15. R. Kodiyath, J. Wang, Z. a. Combs, S. Chang, M. K. Gupta, K. D. Anderson, R. J. C. Brown and V. V. Tsukruk, Small, 2011, 7, 3452–3457.
16. X. Wei and D. K. Roper, J. Electrochem. Soc., 2014, 161, D235–D242.
17. D. Pristinski, S. Tan, M. Erol, H. Du and S. Sukhishvili, J. Raman Spectrosc., 2006, 37, 762–770.
18. W. Lee, R. Scholz, K. Nielsch and U. Gösele, Angew. Chem., Int. Ed., 2005, 44, 6050–6054.
19. Y. Qin, X. Ji, J. Jing, H. Liu, H. Wu and W. Yang, Colloids Surf., A, 2010, 372, 172–176.
20. K. G. Stamplecoskie, J. C. Scaiano, V. S. Tiwari and H. Anis, J. Phys. Chem. C, 2011, 115, 1403–1409.
21. J. Sun, D. Ma, H. Zhang, X. Liu, X. Han, X. Bao, G. Weinberg, N. Pfänder and D. Su, J. Am. Chem. Soc., 2006, 128, 15756–15764.
22. S. L. Lai, J. Y. Guo, V. Petrova, G. Ramanath and L. H. Allen, Phys. Rev. Lett., 1996, 77, 99–102.
23. P. Buffat and J. P. Borel, Phys. Rev. A, 1976, 13, 2287–2298.
24. H. Chen, F. Tian, J. Kanka and H. Du, Appl. Phys. Lett., 2015, 106, 111102.
25. C. W. Bale, E. Bélisle, P. Chartrand, S. A. Decterov, G. Eriksson, K. Hack, I.-H. Jung, Y.-B. Kang, J. Melançon, A. D. Pelton, C. Robelin and S. Petersen, Calphad, 2009, 33, 295–311.
26. E. N. Fuller, P. D. Schettler and J. C. Giddings, Ind. Eng. Chem., 1966, 58, 18–27.
27. W. A. Tucker and L. H. Nelken, in, Handbook of Chemical Property Estimation Methods, McGraw-Hill, New York, 1982.
28. H. C. Graham and H. H. Davis, J. Am. Ceram. Soc., 1971, 54, 89–93.
29. G. R. Holcomb and G. R. St. Pierre, Oxid. Met., 1993, 40, 109–118.
30. Y. Han, R. Lupitskyy, T.-M. Chou, C. M. Stafford, H. Du and S. Sukhishvili, Anal. Chem., 2011, 83, 5873–5880.
31. M. Erol, Y. Han, S. K. Stanley, C. M. Stafford, H. Du and S. Sukhishvili, J. Am. Chem. Soc., 2009, 131, 7480–7481.
32. M. Ramstedt and P. Franklyn, Surf. Interface Anal., 2010, 42, 855–858.
33. Z. Zhao, V. A. V. Rossi, J. P. Baltrus, P. R. Ohodnicki and M. A. Carpenter, J. Phys. Chem. C, 2016, 120, 5020–5032.
34. M. G. Albrecht and J. A. Creighton, J. Am. Chem. Soc., 1977, 99, 5215–5217.

---

Footnote

† Electronic supplementary information (ESI) available: Additional annealing treatments of Ag NPs and corresponding SERS measurements. Mass loss estimation of annealed Ag NPs. SEM images of annealed Ag NPs on planar sapphire substrate. Explanation on “normalized ratio” in XPS results. XPS Auger peaks of Ag NPs before and after annealing. Curve fitting of XPS Ag 3d peaks. See DOI: 10.1039/c6ra17725e

---