1. Introduction

Polymers have been proved to be promising materials for single and multimode optical waveguide based devices [1]. Their interesting properties including low optical loss, transparency in a wide wavelength range, cost effectiveness, processing possibility and compatibility with various substrates and suitability for integration with other devices, have enabled these materials to be used in a variety of applications, i.e., electro-optical devices, sensors, couplers, modulators, and photonic crystals [2–6].

Slot waveguides are nanostructures that confine light at a subwavelength scale in a low refractive index material sandwiched between two narrow high refractive index rails. The field enhancement in these structures is based on the discontinuity of the electric field at the interface between high and low refractive index materials [7]. For that reason, most of the slot waveguides are fabricated in high refractive index materials such as silicon [8–10], in which the slot width should be around 100 nm in the 1.3 and 1.55 µm telecommunication windows. Unfortunately, the sub-micrometer dimensions may limit the application of the slot waveguides. Indeed, some infiltration difficulties may rise when the slot region has to be filled by non-linear materials or fluidic analytes [11]. Lower refractive index based slot waveguides allow a wider slot region and thus better infiltration possibilities. However, the trade-off is a possible leak of light in the rails and a lower slot mode confinement. Nevertheless, theoretical and experimental research results confirm the possibility of realizing these structures in low refractive index materials, such as polymers [12,13]. To achieve a good confinement in the slot region while maintaining a single mode operation, the slot walls have to be as high as possible [13,14]. A high aspect ratio (rail height versus width) leads to demanding fabrication methods that are not always compatible with low cost mass production processes [15,16].

Previously, we have shown theoretically that employing thin layers of high refractive index materials results in an enhanced light confinement in the slot region and relaxed dimensions by increasing the index contrast in the slot waveguide [17]. The thin high refractive index layer on polymer slot pushes the field in the slot area while reducing the field confinement in the polymer rails with lower aspect ratio.

Here we present the realization and experimental demonstration of a layered Young interferometer based on a polymer slot waveguide coated with high refractive index materials fabricated by low cost and mass production compatible methods, i.e., nanoimprinting [18] and atomic layer deposition (ALD) [19].

2. Theoretical design

The proposed Young interferometer includes a polymer slot waveguide coated with a high refractive index bilayer of Al₂O₃/TiO₂. Figure 1 is a schematic view of the Young interferometer showing the cross section of the waveguide in the slot region and the tip of the adiabatic ridge to slot waveguide coupler. The input region of the interferometer contains a 2.5 µm ridge waveguide which is tapered to a 1 µm single mode waveguide.

 

Fig. 1 Schematic view of our Young interferometer with the cross section of the waveguide in the slot waveguide and a tip of the adiabatic strip to slot waveguide coupler.

Download Full Size | PDF

The coupled light is evenly split into two arms spaced 50 µm from each other by a 3 dB Y-branch. In order to achieve a high confinement in the slot region as well as maintaining the monomodal behavior, the width and height of the polymer rails were set to 650 nm and 320 nm, respectively. For this configuration, Ormocore, which is an organic-inorganic polymer with a refractive index of 1.54 at λ = 975 nm, was selected for the polymer slot waveguide. This material has low absorption and a high glass transition temperature which allows the use of the ALD technique without material degradation. The polymer waveguide is covered with a 5 nm thick Al₂O₃ layer of refractive index of 1.63. This thin layer plays a twofold role in our fabrication process. It is a protective layer for polymer rails and an etching stop layer for the TiO₂ layer in etching process. High refractive index (n = 2.34) TiO₂ coatings only on the vertical sides have a thickness of 90 nm. Al₂O₃ and TiO₂ were chosen mainly due to their transparency in the visible and near infrared wavelengths, and due to the possibility to be deposited by ALD at a relatively low process temperature (120 °C). Original slot width of 340 nm reduces to 150 nm after Al₂O₃/TiO₂ bilayer coating.

The mode profile and field confinement shown in Figs. 2(a)-2(d) were calculated using the Fourier Modal Method (FMM) for TE polarization at the wavelength of 975 nm. This wavelength was selected due to 20 times lower water absorption in comparison to 1550 nm [20]. The thin layer of TiO₂ above the polymer rails pushes the modal field towards the over-cladding layer which leads to a reduced field overlap in the slot waveguide. Further increase in TiO₂ layer thickness results in leaky modes in both slot and ridge waveguides [Figs. 2(a) and 2(b)] and enables multimodal behavior as well. However, by applying the TiO₂ layers only to the vertical sides of the polymer rails the mode is well confined in the slot region [Figs. 2(c) and 2(d)] and the electric field is enhanced in the slot. All the calculations have been done by assuming that water with refractive index of 1.327 is a cover medium. Our calculations show that with addition of the high refractive index layers, the mode is strongly localized in the slot region with the width of 150 nm [Fig. 2(d)], which is 50% wider than typical slot waveguides [8–10].

 

Fig. 2 Simulation results of the amplitude of lateral component of the electric field (E_X) of the quasi-TE mode in (a) ridge, (b) slot waveguide with a 90 nm of TiO₂ layer all over the waveguides, with down etched TiO₂ layer in (c) ridge, (d) slot waveguide, and (e) adiabatic strip to slot waveguide coupler.

Download Full Size | PDF

To enable efficient coupling of light from the ridge waveguide to the slot waveguide, the 1 µm wide ridge waveguide is gradually narrowing to the slot rail width [21]. Alongside the narrowing ridge, another slot rail is introduced with a 50 nm tip gradually expanding to the slot rail width. The length of this adiabatic ridge to slot waveguide coupler is 50 µm. The coupling process in the adiabatic coupler was simulated with the Finite Difference Time Domain (FDTD) method. Figure 2(e) shows the lateral electric field (E_x) propagation along the ridge to slot waveguide coupler.

3. Fabrication

The Young interferometer was fabricated by a combination of nanoimprinting and atomic layer deposition, which both are considered as low cost techniques. Since the nanoimprinting process is based on the replication, the quality of the master mold has an important effect on the quality of the final sample. With regard to the small dimensions of the structure, the fabrication of the polymer waveguide started by creating a silicon-HSQ master mold by electron beam lithography. HSQ (Hydrogen silsesquioxane) is a high resolution negative tone resist, which is often used for fabrication of small features (<600 nm) without etching. In addition, it has very small line width fluctuations compared to other e-beam resists [22]. Furthermore, it can be used as a stamp in the nanoimprinting process due to its mechanical properties [23].

The process involves a single patterning step without a plasma etching process for transferring the pattern. Therefore, additional surface roughness, potentially induced from the plasma etching, can be avoided. First, the silicon wafer was spin coated with a single layer of diluted HSQ (DOW corning Fox16) e-beam resist and methyl isobutyl ketone (MIBK) with the ratio of (1:1) to obtain a layer thickness of 320 nm. The resist layer thickness, transferred directly as a height of waveguide, can be controlled by adjusting the spin coating process. The layer was exposed by e-beam lithography (Vistec EBPG 5000 + ES HR) with an area dose of 9000 µC/m² and developed with water diluted Microdeposit 351 developer. After development, the stamp was hardened for 3 hours at 300 °C. Figure 3 shows scanning electron microscope (SEM) images of the different parts of the Young interferometer on the Si-HSQ stamp: the rail walls are vertical and the surfaces are smooth.

 

Fig. 3 SEM images of a) slot waveguide and b) strip to slot waveguide coupler and c) Y- junction on the Si-HSQ stamp.

Download Full Size | PDF

The fabrication process for the polymer waveguide has been illustrated in Fig. 4. Si-HSQ master stamp had the similar structure to the final imprinted sample. Therefore, in order to obtain the negative structure for imprinting, another mold was replicated from the master stamp. For that purpose, the Si-HSQ stamp was treated with an anti-adhesion layer and replicated by dispensing Ormostamp UV curable resist on a fused silica wafer resulting in the final working stamp (WS). The replicated WS was used for fabrication of the polymer slot waveguide by UV nanoimprinting. For fabrication of polymer waveguide, a thin layer of diluted Ormocore-maT1050 (1:11) was coated on an oxidized silicon wafer (3 µm thermal oxide layer) and baked at 120 °C on a hot plate for 10 min to remove the solvent.

 

Fig. 4 Schematic illustration of fabrication process starting from replication of working stamp (WS) from Si-HSQ master stamp until fabrication of the final ALD-coated polymer waveguide structure.

Download Full Size | PDF

The final polymer structure was fabricated by pressing the polymer WS stamp against the Ormocore layer with a thickness of 470 nm with 9 bar pressure and curing for 60 s in Eitre^® Obducat nanoimprinting equipment.

Figure 5(a) shows a SEM image of the cross section of the polymer slot waveguide without high refractive index layers: the aspect ratio in the polymeric rails and slot waveguide is less than one. This facilitates the fabrication of the structure by nanoimprinting. Figure 5(a) also illustrates the fact that because of chosen relaxed dimensions, the polymer rails are in upright position without a rounding error on top of the rails, which would decrease the field confinement in the slot and lead to decreased sensitivity of targeted biosensor [9]. Figure 5(c) depicts the adiabatic coupler with the tip width of 50 nm, which means the structure with an aspect ratio of 6 was fabricated successfully. It should be noted that according to the design of the structure, the reduced aspect ratio allows fabrication of details with the aforementioned resolution by simple nanoimprinting (the aspect ratio in the tip of the coupler part with a similar width in a low refractive index slot waveguide, without applying the TiO₂ layers, would be about 16 which would make the fabrication of the details with this resolution nearly impossible). Figure 5(e) is the SEM image of the 3 dB Y- junction. By comparing Fig. 3 to Figs. 5(a), 5(c), and 5(e), it can be seen that the structure has been precisely replicated. After replication of the polymer waveguide, the structure was coated with a 5 nm Al₂O₃ layer at 120 °C following a 90 nm of TiO₂ layer at the same temperature. Applying the ALD process with this temperature simultaneously hardens the polymer structure as well.

 

Fig. 5 SEM images of (a) cross section of a polymer slot waveguide, (b) cross section of ALD- coated polymer slot waveguide after down etching of the top TiO2 layer, (c) strip to slot waveguide coupler on polymer, (d) down etched ALD-coated strip to slot waveguide coupler and (e) Y- junction on polymer waveguide, and (f) Y- junction in down etched ALD-coated polymer waveguide.

Download Full Size | PDF

Al₂O₃ and TiO₂ layers were grown from TMA (trimethylaluminum) / H₂O and TiCl₄ (titanium tetrachloride) / H₂O precursors, respectively, by the ALD equipment (Beneq TFS 200). By using a fluorine based plasma etching, the TiO₂ layer was further removed from the top of the polymeric rails as well as from the bottom of the slot region. As it can be seen in Figs. 5(b) and 5(d), the layers remain smooth also after the etching process. The gap in the beginning of the Y-junction in polymer [Fig. 5(e)] is reduced to zero after ALD coating [Fig. 5(f)]. The Al₂O₃ layer appears to be a perfect etching stop layer for TiO₂ in fluorine based etching and it also prevents any damage or surface roughness on the polymer rails due to etching. By applying this thin etching stop layer of Al₂O₃, the etching time can be adjusted with relaxed tolerances to achieve a repeatable fabricated structure similar to the design.

It is important to note that ALD is a unique deposition method which can provide a conformal coating with a precise thickness control. The coating thickness can be controlled by the thickness of the grown layer in each cycle, which are 0.069 nm and 0.042 nm for TiO₂ and Al₂O₃ in this work, respectively. Combined with a selective etching between the two layers, the ALD method enables the deposition of TiO₂ on vertical sidewalls of the slot waveguide leading to high confinement of light in the slot. Moreover, the stamp fabrication and pattern replication by the nanoimprint method can be carried out with a low aspect ratio as observable in Fig. 5.

A thick over-cladding of Ormocomp was spin coated on the sample to protect the waveguides from any damage when mounting a fluidic cell on the sample. A window with the length of 5 mm was opened on the sample by UV masking. This window was used for characterizing the sample for a sensor application.

4. Performance of Young interferometer sensor

Light from a wavelength stabilized laser (QFBGLD-980-5, Photonics LLC) emitting at 975 nm was coupled into a 2.5 µm wide input waveguide from tapered polarization maintaining fiber forming a spot about 2.5 µm. A TE polarized state was confirmed with an external polarizer prior to the measurement. The modal intensity distributions and interference patterns were imaged through a 40x microscope objective onto the camera (UI-3240CP-NIR-GL, IDS Imaging Development Systems GmbH). The flow cell was mounted on top of the chip and analyte solutions were pumped on the sample with a syringe pump (Nexus 3000, Chemyx Inc.) at the rate of 100 µL/min. Figure 6(a) shows the modal output of the two arms of the Young interferometer. The phase difference between the two arms due to the presence of the slot waveguide leads to an interferogram which can be observed when the objective was defocused and moved 1.135 mm away from the focus point [Fig. 6(b)].

 

Fig. 6 (a) Output intensity profiles of the ALD-coated Young interferometer of the reference arm (right) and the sensing arm, including an 800 µm long slot waveguide section (left), and (b) interference pattern captured after applying water as a cover medium in the sensing window.

Download Full Size | PDF

In order to investigate the performance of the fabricated Young interferometer for detecting the refractive index changes of aqueous media, ethanol-water solution with different mass concentrations of 0.00178%, 0.0167%, and 0.178% were prepared. In a sensor device, with a Young interferometer as a transducer, the aqueous homogeneous refractive index sensing, which is calculated as $S_h=\frac{\partial n_{\text{eff}}}{\partial n_{\text{c}}}.$, can be measured by measuring the phase shift $\Delta \phi$ in the interference pattern resulting from the change in refractive index of the cover medium $\Delta n_c$ as

The refractive index variation of the cover medium $\Delta n_c$ is the refractive index difference between ethanol-water solutions with different mass concentrations $C_{Eth}$ and DI water which can be calculated as [24]

 

Fig. 7 Measured phase shift after applying an ethanol-water solution with the concentration of a) 0.00178%, b) 0.0167%, and (c) 0. 178%. Red lines are the results after filtering.

Download Full Size | PDF

In the present ALD-coated polymer slot waveguide interferometer, the sensitivity of both arms was increased by applying high refractive index materials, which affects the total result of the device. According to the Eq. (1), the operation of the device for sensing application can be significantly improved by covering the reference arm. For instance, by covering the reference arm, the theoretical phase shift of the device for the ethanol-water solution with a concentration of 0.178%, is increased from 0.07 to 1.3 rad in this device [Fig. 8(a)]. Figure 8(a) also compares the total phase shift of the device against the refractive index change of the cover medium for both experimental and simulated results done by both FMM and OptiFDTD commercial software. The nonlinearity of the experimental results can be due to accumulation of ethanol on the surface of the device in the sensing window due to presence of the TiO₂ layers. In order to confirm this assumption, we simulated the slot and ridge waveguide with a 3 nm of the additional layer with refractive index of the 1.5 on the sides of the rails, mimicking the adsorption of ethanol by the TiO₂ layers. According to our simulations with FMM, for the ethanol-water solution with the concentration of 0.178% the phase shift decreases to 0.049 rad.

 

Fig. 8 a) The comparison of the experimental and the theoretical values of phase shift calculated by FMM and Opti FDTD. The theoretical phase shift for the same device with the covered reference arm was calculated by FMM, and b) effective refractive index change of the ALD-coated polymer slot waveguide interferometer against the change in the refractive index of the cover medium. The slope of the linear fitting gives the homogeneous sensitivity $S_h$ of the device based on the measured values.

Download Full Size | PDF

In order to calculate the homogeneous sensitivity $S_h$, the total effective index change of the device ${\left(\Delta n_{\text{eff}}\right)}_{total}$ was calculated by replacing $\Delta \phi$ with the measured values and assuming L = 800 µm and λ = 975 nm in Eq. (1). The homogeneous sensitivity of the device is the slope of the linear fitting for the calculated values of ${\left(\Delta n_{\text{eff}}\right)}_{total}$ versus $\Delta n_c$ related to different concentrations of the ethanol-water solution [Fig. 8(b)]. Therefore, the homogeneous sensitivity $S_h$ of the device is obtained as 0.051.

5. Conclusion

A Young interferometer based on ALD-coated nanoimprint replicated slot waveguide structure was fabricated. Despite its prior complicated design, the fabrication remains reliable, low cost and repeatable. Applying the thin layer of Al₂O₃ avoids any roughness or damage of the polymer structure after the etching process. The thin vertical TiO₂ layers on sides of the polymer slot waveguide increase the confinement in the slot waveguide with a small footprint and a low aspect ratio. Such a wide slot gives more space for infiltration of liquids and the relaxed dimensions simplify the fabrication of the smallest details by nanoimprinting. The device was characterized at 975 nm wavelength and the feasibility of its functionality for aqueous refractive index sensing at the same wavelength was experimentally demonstrated. A refractive index change of 1 × 10⁻⁶ RIU, with a sensing length of 800 µm, was measured. The sensitivity of the device was measured to be 0.051. It was shown that the fabricated low cost structure can be applied for bulk homogeneous sensing.