1. Introduction

The ion-exchange technique has been used for centuries to modify optical absorption properties of glasses for purposes of decoration and coloring [1]. However, only in the late 70´s with the pioneering work of Izawa and Nakagome [2] and Gialorenzzi et al [3], it was showed that the technique is suitable to produce optical waveguides [1]. Since then, this fabrication process has been used in several applications of integrated optics: optical communications, signal processing and optical sensing [4,5].

The advantages of the technique are low material and processing costs, low optical losses produced in the waveguide and compatibility with optical fiber assembly. Typically, in this procedure, Na⁺ ions present in the glass are exchanged with other ions from a melt salt. These ions have different ionic radius and polarizability from those of Na⁺, inducing stress in the glass matrix, accompanied by modification of its refractive index [1]. The most common ion used is silver (Ag⁺) due to its great variation in the refractive index compared to other ions, such as potassium (K⁺) and thallium (Tl⁺) [6].

Moreover, the ion-exchange with silver gives rise to an intense photoluminescence (PL) emission, which is potentially useful for optoelectronic applications [7]. For that reason, many works have been devoted to increase the photoluminescence response, where post annealing and ion bombarding are regularly used after the ion-exchange, forming metal NPs on the bulk volume [8,9]. The study of PL in silver ion-exchange waveguides also allows clarifying the mechanism of structural changes in glasses. Among these changes, one that should be cited is the migration of Ag⁺ towards the surface by post annealing process under hydrogen atmosphere [10,11]. The increase in the Ag⁺ PL intensity was the evidence for the precipitation of metallic silver nanoclusters in near surface glass. However, this migration effect of silver emerging onto the surface is barely explored and not even characterized in literature.

On the other hand, Ag⁺ ions are extremely reactive when exposed to different environments, such as air, moisture and ozone or sulfur atmospheres [12]. In this work, an experiment was performed on bare silver films with 99.9% purity exposed to different atmospheres. The conclusion was that the principal route to silver corrosion at the surface was caused by ozone, which generates reactive atomic oxygen that reacts rapidly with Ag to form Ag₂O.

Furthermore, Ag₂O systems can also produce large SERS signals from photoinduced Ag cluster formation [13], which means that is possible to photoreduce the oxide surface to yield a corrugated metallic silver substrate. SERS in silver glasses system was also reported in [14] by a two-step ion exchange followed by etching of the substrate to expose the silver NP’s. Reduction processes can also be obtained simply using an ordinary heat treatment, as showed by Libardi and Grieneisen [15]. Therefore, the ability to modify silver/silver oxide systems allows the development of a new template for integrated optical devices through local heating using a laser pulse, for example. This has been demonstrated when an extremely high heating rate was applied in a particular region of the material, enabling design of structures such as a metal gratings in silver oxide films [16]. The ability to structure metal nanoparticles and metal-oxide surfaces is of considerable interest because of its potential application in advanced plasmonic-based devices, e.g., surface plasmon amplification by stimulated emission of radiation (SPASERS) [17]

For Ag-doped planar waveguides produced by ion-exchange, the silver NPs on the waveguide surface can interact with the environment causing changes in its topology and optical properties. Particularly, the formation of silver nanoclusters in glass underneath the surface, under post annealing treatments, was previously reported in the literature [10,11]. However, so far as we know, the occurrence of this effect in room temperature under air atmosphere, whereas the nucleation of silver NPs occurs outside the glass matrix, has not been reported yet.

In this context, the aim of this work was to investigate through Spectrophotometry, Spectral Ellipsometry (SE), AFM and Raman measurements this process of migration and nucleation of silver NPs at the surface of ion-exchange silver waveguides at room temperature. Finally, to demonstrate the viability to use the layer formed in the process as a new template to integrated devices, results of two distinct types of applications are presented. First, it is evidenced the amplification of the Raman signal of Rhodamine 6G (Rh6G) deposited on the surface of the formed layer by the SERS effect. Next, a micrometric holographic diffraction grating was fabricated using a laser beam heating and it was subsequently characterized.

2. Experimental

2.1 Sample preparation

Soda-lime glass slides Knittel Optifloat of 3 x 1 inches [SiO₂ (72 – 73%), Al₂O₃ (0.5 – 0.7%), Fe₂O₃ (0.10 – 0.13%), RO(CaO + MgO) (12.7 – 13.1%) and R₂O (Na₂O + K₂O) (13.2 – 13.6%)] were cleaned using two-step RCA routine to remove metals and organics: H₂SO₄ + H₂O₂ (4:1) followed by H₂O + NH₄OH + H₂O₂ (4:1:1). Afterwards, the slides were washed with deionized water and dried with N₂ flow.

Three sets of waveguides (two identical samples by set) were prepared by the ion-exchange process [1], using a melt salt of NaNO₃ and AgNO₃ (5% molar) in a furnace at 350 ± 2 °C during 5, 30 and 120 minutes, which were labeled as samples 05, 30 and 120, respectively. After quenching, samples were cleaned again with distilled water to remove the salt residues and dried with N₂.

2.2 Aging process and characterization techniques

Spectrophotometric measurements were performed with a CARY 5000. The absorbance caused by the silver doping and by the aging in the waveguides was assessed, subtracting bare glass contribution, from transmittance (%T) and total reflectance (%R) measurements in the wavelength range of 300 to 700 nm.

SE measurements were performed using an Ellipsometer SOPRA GES-5E with a microspot accessory to collect light reflected only from the sample surface (incident angle = 68° and wavelength range of 300 to 700 nm). The sample analysis and the determination of their optical constants were performed by the Ellipsometer analysis software (WinElli).

AFM measurements were performed using a microscope DI SPM IIIa from Digital Instruments, with resolution of 4 nm and 2.0 x 2.0 µm of area scanning. The results were used to confirm topological changes on the surface of the samples.

In order to evidence the SERS activity, Raman measurements were performed using a Horiba Jobin HR320 spectrometer connected to an Olympus microscope. The excitation was provided by the 632.8 nm line of 10 mW HeNe laser focused in a spot of approximately 2 µm. The scattered light was collected in back reflection geometry and filtered with a Super-Notch Plus filter (HSNF 633-1.0, Kaiser Optical Systems, Inc).

2.3 Holographic diffraction grating fabrication

The holographic diffraction gratings were fabricated using the 355 nm beam of a Q-Switched Nd:YAG laser from Quantel (model YG780) operating in single shot mode with 2.2 mJ by pulse. Higher pulse energies were avoided to prevent the layer ablation. The interference pattern formed was created over the sample surface passing the 10 mm waist of laser beam by a BK-7 Fresnel bi-prism (roof angle = 2°). The HDG formation was monitored in situ by the intensity measurement of the diffracted first order in reflection of a continuous probe laser (532 nm and 35mW) collected on a photodiode. HDG efficiency was also evaluated based on this signal.

3. Results and discussion

Figure 1(a) shows the coloration changes in each waveguide for the samples 05, 30 and 120 after 35 days of air exposure, as well as the waveguide after the removal of the layer by a soft scrub with distilled water (labeled as “cleaned region”). Their absorbance spectra, discounting bare substrate, are presented in the Figs. 1(b), 1(c) and 1(d) respectively for the cases: as exchanged (labeled “new”); 5, 15, 20, 25 and 35 days of aging; and after the cleaning of the waveguide surface (labeled “clean”).

 

Fig. 1 (a) Picture of the samples after 35 days of aging and, in comparison, a region cleaned by a soft scrub with distilled water. Also are shown the ΔAbs spectra (bare glass discounted) from Samples 05 (b), 30 (c) 120 (d) for the following cases: waveguide as exchanged (labeled “new”), aged and after it has been cleaned (labeled “clean”).

Download Full Size | PDF

The doping with Ag⁺ ions created a characteristic absorption peak around the 315 nm (3.94 eV). Lawless et al. [18] studied the reduction and aggregation of silver ions in colloidal silica and observed the increase of a similar 315 nm absorption peak in their experiments, which was attributed to the formation of Ag₂⁺ cations on the surface of SiO₂ particles. The authors concluded that the most probable mechanism to produce these molecular cations was the reaction between Ag⁺ cations, previously adsorbed by SiO₂ particles.

However, the most significant modification in the absorbance spectrum was observed around the wavelength of 450 nm (2.76 eV), where a very large absorption band appears. Al-Kuhaili [19], in his work about the characterization of silver oxide produced by thermal deposition, showed that, particularly, Ag₂O has a large absorption peak for energies between 2.5 and 3.1 eV, which corresponds to the wavelength region where the second absorption band is observed. This indicates that the Ag⁺ cations on the waveguide surface react with air atmosphere forming Ag₂O. Nevertheless, since the Ag₂O absorption band is expect to be at 550 nm [20] and the surface plasmon polariton (SPP) resonance for silver in ion-exchange samples is around 410 nm, it is possible to affirm that the absorption band observed in 450 nm is a combination from these two characteristic bands. After cleaning the waveguide surface, the peak at 315 nm reduces its amplitude in all samples, whereas the band peak around 450 nm disappears entirely (spectrum labeled “clean”). In fact, this absorbance spectrum presents a shape similar to that observed for the brand new sample, only shifted to lower values of absorbance, indicating that the silver concentration in the waveguide was reduced after the layer removal.

The same samples were also characterized using SE. The dispersion curves of the bare soda lime glass were calculated directly by the ellipsometer analysis software and presented in the Fig. 2(a) , since for bulk samples there are analytical solutions for n(λ) and k(λ), as described in [21].

 

Fig. 2 Dispersion curves: (a) from soda-lime glass, Ag-doped waveguide surface as exchanged and cleaned after 35 days of aging (sample 120); (b) from the layer formed over waveguide surface (10, 20 and 35 days of aging).

Download Full Size | PDF

The processes of ion-exchange, followed by the diffusion of Ag⁺ ions inside the glass, are responsible for the waveguide formation due to the increase of the real part of the glass refractive index values, n(λ), in this region. Usually, this refractive index has a smooth profile with depth, ranging from a maximum value, localized in the region close to the waveguide-air interface, to the value measured for bare glass [22]. In this case, it is reasonable to assume that the graded index waveguide has a similar behavior as a bulk material, where the light reflected, which is collected and analyzed by the ellipsometer, corresponds to the waveguide surface region. This was demonstrated in previous work, where the refractive index from the waveguide surface was obtained directly through polarimetric measurements [23].

Based on this model, the refractive index values (n and k) of the waveguide surface were calculated for all samples and their dispersion curves presented values and shapes similar to those shown in Fig. 2(a) for sample 120 (as exchanged). In comparison with bare glass, the real refractive index increased by Ag⁺ doping reaching values around 1.6, which agrees with values reported in the literature [22]. Furthermore, the imaginary refractive index (k) also increased, which was caused by two factors: the light scattering caused by the stress in glass matrix due the inclusion of Ag⁺ (diameter ≈0.252 nm) that is a larger ion than Na⁺ (diameter ≈0.19 nm) and also due light absorption, as can be seen by its maximum value around 315 nm, indicating the presence of an absorption peak, which agrees with the absorbance spectra obtained from the spectrophotometry measurements.

Due to the interaction with the atmospheric air, a layer starts to grow over the waveguide surface, so that bulk model used before is no longer applicable here. The new model assumes an equivalent effective medium formed by the growth of an isotropic and continuous layer above the waveguide surface. Using this model, a good fit agreement was obtained for samples with 10 days or more of aging. The shape and evolution of the refractive indices with time are similar to those shown in Fig. 2(b) for sample 120.

The dispersion curves presented values compatible with a layer formed by silver oxide (Ag₂O) as reported in the literature [24]. Particularly, the imaginary part indicates the presence of a great absorption band that has a good agreement with the absorbance spectra previously obtained from the spectrophotometry measurements. Moreover, the analysis using SE also gives information about the layer physical thickness, which reaches, after 35 days of aging, values of 1.1 nm, 2.2 nm and 11.8 nm for the samples 05, 30 and 120, respectively.

In the case where the layer was removed, the model used in the analysis was the same as for bare glass and the refractive indices are displayed at Fig. 2(a) for sample 120 (labeled “sample cleaned”). The behavior for all samples was very similar: values of the real refractive index were significantly reduced at the waveguide surface, which can be explained by the migration of silver to form this layer, reducing the silver concentration and creating void spaces in the waveguide surface region. In comparison, for bare glass, the surface region was originally filled by Na⁺ ions that were exchanged with Ag⁺ ions to form the planar waveguide (sample as exchanged). This hypothesis also can explain the behavior of the imaginary part of the refractive index, which increases significantly in the cleaned samples when compared with bare glass and also with the sample as exchanged due the presence of these void spaces that increase the light extinction. However, for the wavelength region under 500 nm, the shape of k(λ) curve changes in comparison with the sample as exchanged, presenting low values for the imaginary refractive index. Observing the Ag⁺ absorption curve in Fig. 1(d) (sample as exchanged), this absorption is present only for light wavelengths below 500 nm. So, even with the increasing of scattering caused by the voids after Ag⁺ migration, the absorption contribution for the imaginary refractive index is significantly reduced in this wavelength region, explaining the change in the shape of k(λ).

Finally, in order to observe changes in the topology of the waveguide surface, AFM images were taken from sample 120, as displayed in Fig. 3 : (a) waveguide as exchanged, (b) 35 days of aging and (c) cleaned.

 

Fig. 3 AFM images of the Sample 120 (2.0 µm x 2.0 µm): (a) waveguide as exchanged, (b) 35 days of aging and (c) cleaned.

Download Full Size | PDF

Comparing images (a) and (b), it is clear that the surface of the aged sample (35 days) is completely different from that of the waveguide as exchanged, resembling only the smooth surface of a glass substrate. Structures with estimated diameters from 30 to 40 nm cover entirely the waveguide surface, forming the continuous layer observed in the picture of aged samples (Fig. 1(a)). After removal of this layer by a soft scrub with distilled water, the structures disappear entirely, as can be seen in image (c), which agrees with the optical measurements.

The structures observed in Fig. 3(b) indicate the aggregation of silver NPs forming clusters on waveguide surface, while the refractive indices obtained by SE agree with values obtained in literature for Ag₂O films [24], the most stable silver oxide, which also presents an absorption band similar to those obtained in the spectrophotometry measurements. In fact, the optical response observed can be assigned to a combination of the effects of silver NPs aggregation and oxidation.

Furthermore, the AFM images also indicate that a migration of the silver ions from inside the waveguide to its surface at room temperature occurred. This is confirmed by the optical measurements in the cleaned samples: reduction in the n(λ) values and increase of light extinction, k(λ), at the waveguide surface obtained by SE, as well as by the amplitude decreasing observed for the 315 nm peak in the absorption graphs obtained from spectrophotometry, both show that silver concentration on the outer waveguide surface decreased after aging.

Similar behavior was observed by Garcia et al. in silica samples containing atomic silver prepared by the sol-gel technique [25]. They showed that a migration of atomic silver and its accumulation at the film-substrate interface occurred, followed by a slow silver aggregation into larger clusters. The entire process is slow at room temperature for the silica matrix, taking several months.

In addition to usual planar waveguide application, the presented Ag ion-exchange glass introduces a new sort of plasmonic template, which has great potential as a new candidate for integrated devices, e.g., as a plain substrate for SERS experiments.

To illustrate this, two different approaches were evaluated: first, characterizing the plasmonic nature of this developed material and second, the ability to process the material as an optical dielectric-metal switchable nanometric template, allowing the construction of diffraction gratings.

Since the aging process creates a layer over the waveguide surface and this layer is formed by a mixture of silver oxide and silver NPs, as evidenced by the UV-VIS absorption spectra that shows a superposition of SPP resonances and Ag₂O band, it is reasonable to expect that a significant SERS signal can be detected. Besides, the nanometric structures, evidenced by the AFM at formed layer surface, resemble aggregate colloids known as “hotspots”, which additionally enhance SERS through the structural factor of local field enhancement. Within this approach, SERS has proved to overcome the disadvantage of the small cross section of Raman spectroscopy, allowing the single molecule Raman spectroscopy [26,27].

To confirm the SERS effect, the amplification of the Raman signal of an organic molecule adsorbed on these templates was tested. The samples were prepared by their immersion into the solution of Rh6G (45 µM) in ethanol, followed by solvent evaporation. The Raman measurements were made on the glass used to fabricate the waveguide, as well as on sample 120 after 35 days of aging and on a similar sample, in which was performed a heat treatment to reduce the silver oxide in metallic silver (20 minutes at 80 °C) as showed in Fig. 4 As expected, a SERS signal was observed in aged sample, indicating the presence of silver NPs in the formed layer at waveguide surface. This signal was further amplified, significantly, when the sample was heated, reducing the remaining silver oxide in the layer to metallic silver.

 

Fig. 4 (a) SERS signal collected from Raman measurements of Rh6G for bare glass (black line) compared to sample 120 after 35 days of aging (blue line) and after to be heated at 80 °C for 20 min. (red line) (spectra are presented with identical scales, however they are shifted for a better comparison).

Download Full Size | PDF

Furthermore, the simplicity to fabricate and processing this material is a great advantage over similar templates used for SERS. Simo et al. [28] produced silver NPs inside of soda-lime silicate glass through a combined thermal and chemical three-step methodology. First, the ion-exchange process was performed to obtain metal atoms and smaller clusters in the vicinity of the glass surface. This is followed by a thermal reduction in H₂ atmosphere at 500 °C and an annealing at 600 °C in air atmosphere to promote the metal particle growth. However, as mentioned by the authors, for SERS investigation these metal particles need to be uncovered using a mechanical or chemical process at glass surface.

A second application of the silver/silver oxide layer was based on the development of HDG by laser thermal modification. Diffraction gratings with period of 57 μm were produced in sample 120 using the interference pattern created by a Fresnel biprism. As a result, the laser beam has selectively heated the sample, producing a refractive index periodicity in the uppermost nanometric layer of the metallic oxide. This periodicity corresponds to alternating regions where the silver oxide was reduced to metallic silver (illuminated regions), with regions that remain unchanged. The experimental setup and a picture of the fabricated grating, obtained from an optical microscope, are showed in Fig. 5 .

 

Fig. 5 Setup used to fabricate the holographic diffraction gratings: a NdYAG laser beam passing through a Fresnel bi-prism creating an interference pattern with period of 57µm over the sample surface. As result, a HDG is produced in the layer surface as can be seen in the picture obtained from a Microscope (magnification of 50x). Three Nd:YAG laser pulses of 2.2 mJ were used to heat the sample and transform the silver oxide into metallic silver (clear regions in the picture).

Download Full Size | PDF

The measurements with the probe laser confirmed the period of 57 μm for the HDG and its maximum diffraction efficiency (around 0.5% of the total reflected power) was attained after exposing the material to three laser pulses. The ability to produce gratings in this material allows designing them to specific applications, e.g., coupling light in the ion-exchange planar waveguide localized above.

On the other hand, only the light processing of the material could be used to produce a metal surface with nanometric roughness to enhance PL signal, since it is know from literature that luminescence might be enhanced by rough metal films, islands, nanoclusters and nanoparticles when compared to bulk metal [29].

In order to control the template fabrication, the parameters for the layer growth and aging process still need to be investigated and completely determined. A study that describes the Ag profile evolution within the bulk and relates it to diffusion through the interface is underway. However, the aging process is also closely related to the diffusion and until now we were not able to obtain a clear understanding of this process kinetic.

4. Conclusion

In this work, an investigation through optical and AFM measurements showed a migration of silver Ag⁺ cations from inside of the waveguide to its surface at room temperature. These cations interact with the air atmosphere, forming a layer by the growth of silver NP clusters and silver oxide (Ag₂O) over the waveguide surface. As a result, a reduction of silver concentration was observed in the region close to the waveguide surface. The simplicity described for the production and straightforward processing indicate that such a model structure, aggregating waveguiding, fluorescence and plasmonic enhancement, turns out to be a simple breadboard for photonic integration. The proposed applications to this new plasmonic template are herewith justified by the measurement of Rh6G SERS on the film surface, as well as by the photothermal fabrication of micrometric metal-oxide holographic diffraction gratings.