1. Introduction

In the past decade, tremendous works have been devoted to making proper materials and structures for achieving ultrafast and ultra-low power integrated all-optical switches and modulators at the nanoscale [1–5]. Photonic crystal (PC) structures based on semiconductor materials are considered to be one of the most excellent candidates in fabricating nanoscale integrated all-optical circuits. However, the relatively weak and slow nonlinear optical properties of pure semiconductor materials have hindered the further progress in reducing the consumption power or pulse energy and the response time. So on one side sophisticated methods such as ion-implantation or placement of reverse bias are adopted to improve the nonlinear performance of semiconductor photonic devices [6,7]. On the other side, people also resort to new kind of materials with large Kerr nonlinearity and fast relaxation time, such as pure polymers or nanocomposite doped polymer materials [4,8].

A straightforward and efficient way for fabricating high-performance nonlinear photonic crystal (NPC) integrated optical devices is to combine planar semiconductor PC structures with these highly nonlinear polymer materials. The key point in fabricating this semiconductor-polymer compound NPC lies in how to uniformly and densely fill polymers into the small air holes with a diameter of only hundreds nanometer. Previously, several techniques have been employed for the infiltration of PC with organic composites [9–16]. However, specific problems, such as delamination and shrinkage of the infiltrants, have to be addressed in these technical processes. For example, solution infiltration of PC with polymers results in shrinkage and trapped air bubbles after the removal of solvent [10,11]. Monomer infiltration and subsequent polymerization can lead to a good degree of infiltration, nevertheless, the materials suitable for this approach are limited and oftentimes they experience 10%–15% of shrinkage after the thermally initiated polymerization [9,12]. Quite recently the vapor deposition method is chosen to completely fill the silicon slot waveguide with specific small organic molecule, where good homogeneity originates from the assembly of an amorphous phase without the formation of microcrystals or grains [17]. So in order to homogeneously fill the hundred nanometer size holes or slots, the above infiltration methods are all limited to certain kind of organic molecules. A more general method, which can be applicable for most polymer materials, especially the nanocomposite polymer materials, is still lacking.

In this work, we present a general and practical technique for fabrication of NPCs by infiltrating polymer into semiconductor PC air hole arrays by utilizing the nano-imprint lithographic (NIL) technique, which is believed to be universal for all the thermoplastic materials. Not only complete compound NPC structures but also compound NPC microcavities have been successfully fabricated with high quality. The quality of the infiltration has been characterized by the cross-sectional scanning electron microscopy (SEM) analyses and optical transmission measurements, both of which confirm the success of this NIL based synthesis method.

2. Compound NPC fabricated by NIL technique

The resulting NPC structure is a multilayer hybrid slab, which is schematically illustrated in Fig. 1(a) . This hybrid nonlinear structure has been proposed and analyzed theoretically in our previous work, where the band gap and cavity mode shift under external pump light were investigated [18,19]. In particular we choose polystyrene (PS) and silicon on insulator (SOI) to build the compound NPC structure. The SOI structure is widely investigated for constructing integrated photonic devices and polystyrene (PS) we chose has large Kerr nonlinearity (the third-order susceptibility ${\chi }^{(3)}$of polystyrene is 1.15 × 10⁻¹² cm²/W) and very fast optical response speed (up to a few femto-seconds) [4,5].

 

Fig. 1 Schematics of the fabricated compound structure (a) and the fabrication procedure (b). The fabricated compound structure is a multilayer structure, from top to down corresponding to polystyrene layer, silicon layer with photonic crystal hole array filled with polystyrene, the silicon dioxide layer and the silicon substrate.

Download Full Size | PDF

The NIL technique was invented by Stephen Y. Chou in 1995 to replicate the nano structure conveniently with the advantage of cost-effective and high throughput [20,21]. For the common thermal NIL the principal procedure is as follows. A mold with nanostructures on its surface is pressed into a thin resist coated on a substrate. Keeping constant high pressure and the temperature beyond the glass transition temperature T_g of the resist, the resist becomes viscous and the flowing of it under high pressure will infiltrate the void region of the nanostructure completely after adequate time. After the removal of the mold the nanostructures are transferred into the resist film. So high-quality polymeric PC slab structures have been successfully fabricated by this technique [22,23]. By now the NIL is all treated only as a lithography technique for generating nano-pattern in the resist film. The great potential as an infiltration technique is largely neglected. So we envisage that if we remove the substrate instead of the mold after the imprint process, we would get perfectly mold-resist compound structures. This is our main idea for fabricating semiconductor-polymer compound NPC based on NIL. In our case, the corresponding mold and resist we used are SOI and polystyrene, respectively. As a matter of fact the NIL based method is expected to cover most of the polymer or nanocomposite polymer and semiconductor materials.

The fabrication procedure is illustrated in Fig. 1(b). Concisely it can be divided into two parts, i.e. the fabrication of the PC mold on SOI structure and the preparation of the PS film coated on the passivated silicon (Si) substrate. In our case silicon PC patterns were fabricated using Electron-Beam Lithography (Raith150, Germany) combined with HBr-based Inductively Coupled Plasma (System100, Oxford instruments, UK) etching. Then the NIL (Eitre-3, Obducat, Sweden) technique was adopted to fill the air holes array of silicon PC slab with polystyrene for the formation of compound NPC structure. The final fabricated silicon PCs, which serve as mold, consist of triangular lattice array of air holes etching through the top silicon layer of the SOI wafers with lattice constant as 400 nm and diameter as 252 nm, which is shown in Fig. 2 . The thickness of the top silicon layer is 235 nm. Before the preparation of the PS film, the surface of the Si substrate is treated with 1H,1H,2H,2H-Perfluorooctyltrichlorosilane, resulting in an anti-stick property between the polymer film and substrate surface so as to detach PS film from the substrate during the demold process. Here it must be emphasized that stable temperature and homogeneous large-area pressure during the imprint process are the crucial conditions for the success of NIL. In our NIL system the pressure applied to the mold and substrate are offered by the compressed air, which results in uniform pressure upon the whole area. Stable and program-controllable temperature is provided in the imprint chamber. Under an appropriate imprint condition of temperature, pressure, and contact time, a PS film with thickness of 1-2 μm is left on top of the silicon layer. The optimal imprint condition we found for our process as 160 °C for temperature and 50 barr for pressure with a contact time of 5 minutes beyond the T_g temperature. According to early theoretical work [18], as the thickness of the PS layer above the semiconductor membrane is far larger than the lattice constant, the band location and band gap width of the fabricated compound NPC will reach a steady magnitude that is close to the values for the NPC with an infinitely thick PS overlayer. Furthermore, as the PS film is thick enough, the influence of the roughness of the top surface of the PS on the transmission spectrum can also be neglected.

 

Fig. 2 SEM top view pictures of a triangular-lattice complete photonic crystal. Strip waveguides are visible at the left and right of the PC region. (a) Before infiltration; (b) After infiltration by utilizing NIL technique. The image contrast for the compound structure is low because of the poor conductivity of polystyrene covering the silicon membrane.

Download Full Size | PDF

In order to characterize the effect of the infiltration, we cut the structure using the Focused Ion Beam (FIB). In this process, the 1-2 μm polystyrene thin film has been largely polished out to a very thin and rough layer (tens nanometers thick) covering the silicon structure. Figure 3(a) and Fig. 3(b) show the cross-sectional SEM view of the compound NPC slabs. The residual polystyrene gums can be clearly seen here. We can see that complete filling for all holes is achieved using this technique. The polystyrene fills the holes ideally and conformally without shrinkage and delamination from the side walls. To further clarify that the holes are undoubtedly infiltrated with PS rather than imaging error arising from the poor conductivity of PS, we use a local deposition method to examine whether the holes are infiltrated or not, as is shown in the Fig. 3(c) and Fig. 3(d). Firstly we remove the covered PS layer of the compound structure totally by utilizing of O₂ Reactive Ion Etching (RIE) until the silicon surface emerges exactly. Then we locally deposit a layer of platinum (Pt) by FIB induced deposition. After that we cut the deposited region by FIB, and we can see from the cross-sectional image [Fig. 3(c)] that Pt could not deposit into the holes array. For comparison, we deposit Pt layer on a PC structure without PS imprinted and we can see from Fig. 3(d) that Pt can fill air hole arrays completely. Based on the above SEM analysis results, it can be inferred that the silicon PCs are homogeneously and completely filled with PS.

 

Fig. 3 (a) and (b) are the cross-section SEM image of the compound photonic crystal structure. The dark grey color corresponds to the polystyrene while the light grey color corresponds to the silicon owing to the difference between their conductivity. We can see that polystyrene fills the holes ideally and conformally without shrinkage and delamination from the side walls. (c) The cross-section SEM image of the compound photonic crystal structure after the deposition of Pt layer. We deposit a layer of Pt by FIB induced deposition on the surface of the compound structure after taking out the surface polymer by the O₂ Reactive-Ion-Etching, and we can see that Pt cannot deposit into the holes array, indicating complete infiltration of polystyrene into the air holes. (d) The cross-section SEM image of the non-infiltrated photonic crystal structure after the deposition of Pt layer. In contrast with panel (c), we can see that the Pt can deposit into the air holes and reach their bottom. The bright rectangular bump on top of the slab is the deposited Pt layer.

Download Full Size | PDF

3. Optical characterization of compound NPC

As we know, infiltration of silicon PCs with high index polymer, which leads to an increase of the effective refractive index, will result in the red-shift of the photonic band gap (PBG). To optically characterize the band shift effect of the PBG by infiltration of PS, the transmission spectrum is measured along the Γ-K direction of the NPC, as is represented in Fig. 2. Optical measurements are performed with the end-face fiber coupling system [24–26]. The input optical signal with TE polarization comes from a continuous wave tunable diode laser with wavelength ranging from 1500 to 1640 nm, and launched into one facet of the silicon strip waveguide via a single-mode lensed fiber. The transmission signal is detected by a power meter via another lensed fiber collecting the emitted signal from the output side. The output strip waveguide was gradually tapered from 10 μm at the place adjacent the NPC structure to 1.5 μm at the end face of the whole sample that connects with tapered fiber lens, so that we can collect more signal light transmitting through the NPC structures.

The measured transmission spectra of the complete silicon PC structure before and after infiltration are presented in Fig. 4(a) . In both cases, the transmission contrast between the pass-band and the forbidden-band of the PC is around 15-20 dB, and moreover, the span of wavelength from PBG to pass band takes only 5 nm. It can be observed that the long wavelength band-edge of the PCs exhibits a 20 nm red-shift after infiltration. Furthermore, the transmission contrast of the infiltrated samples is about 5 dB larger than that of the sample before infiltration. This is caused by the increase of the effective refractive index of the structure region after PS infiltration, which leads to the reduction of the out-of plane loss [6]. Figure 4(b) is the calculated transmission spectrum for the empty and the filled PCs simulated by three-dimensional finite-difference time-domain (3D-FDTD) method [27]. Compared with Fig. 4(a) we can see that the experiment curves are 20 dB lower than the simulation results, and this is due to the coupling loss at the two interfaces between the lensed fiber and the input and output Si strip waveguides. The simulation data show an average 20 nm red-shift of the band edge after infiltration of polystyrene, which is close to the value found in experiments. The significant shift of band-edge and large transmission contrast indicate that the filling of the PC holes is complete and uniform. On the other hand, the absolute position and the line shape (e.g., the steepness) of the band edge show a slight deviation between experiment and theory, and this can be attributed to the nanofabrication imprecision as well as the departure of the models used in the numerical simulation from the practical samples under experimental studies. For instance, the input and output ridge waveguides have been neglected in the numerical simulations for the sake of simplicity.

 

Fig. 4 Transmission spectra of the TE-like mode along the Γ-K direction of the empty and infiltrated triangular-lattice PCs. (a) Experimental data, where both of the samples before and after infiltration are measured twice to clarify the validity of measurement data. The two lines with solid cubic symbol are the transmission curves before infiltration, and the two lines with hollow circle symbol are the transmission curves after infiltration; (b) 3D-FDTD simulation data.

Download Full Size | PDF

PC microcavities are considered to be more sensitive to the external excitation power or pulse energy than the complete PC and show great potential in studying the interaction between light and matter [18,28–35]. Hence, we also realize a compound NPC micocavity following the same fabrication procedure to further verify the validity of this NIL method. The structure is based on our previous design [18]. Figure 5(a) is the top-view SEM image of the silicon PC microcavity before infiltration while the SEM image after filling with PS is shown in Fig. 5(b). The left and right sides of the PC microcavity are PC W1 waveguides by removing a row of air holes designed to couple light in and out of the cavity region, as is represented in Fig. 5(a). The lattice constant of the PC, the radius of the regular air hole, the thickness of the top silicon layer and the radius of center defect hole are 420 nm, 124 nm, 220 nm, and 335 nm, respectively. As the thickness of the SiO₂ substrate and the polystyrene cladding layer are greater than the lattice constant, we can consider them as infinite.

 

Fig. 5 (a) Top-view SEM images of PC microcavity before infiltration. Two PC W1 waveguide formed by removing a row of air holes are connected with the cavity region; (b) Top-view SEM images of PC microcavity after infiltration. The image contrast for the compound structure is low because of the poor conductivity of polystyrene covering the silicon membrane.

Download Full Size | PDF

The transmission spectrum of the Si-PS compound NPC microcavity is presented in Fig. 6 . In Fig. 6(a), the black dot curve is the measured data and the red line represents the fitting line of Lorentzian line shape. The loaded quality factor (Q value) can be evaluated from the fitting line and is the center wavelength divided by its full-width at half maximum, which is approximately 227. Figure 6(b) is the theoretical result simulated by 3D-FDTD and the extracted Q value is about 280. The measurement value is a little lower than the theoretical one, and this deviation is mainly induced by fabrication imperfection of the silicon PCstructures instead of the inhomogeneity of the infiltration of PS. In our previous theoretical paper [18], the microcavity in Fig. 6 is designed to be Kerr-effect sensitive because of the on-resonance optical field energy concentrated in the center PS defect hole. The cavity mode resonant frequency of this structure can shift 8.4 nm under the excitation of peak intensity of 80 GW/cm² [2,18]. This value is far larger than the shift magnitude of the ordinary M1 cavity or L3 cavity that is fabricated in the pure air-bridged silicon slab without PS, which is less than 1 nm [3]. In our opinion there are still ways to improve the quality factor such as reducing the coupling loss between the PC waveguide and the cavity, optimizing the cavity structure, and enhancing the micro-fabrication precision and so on. Although the Q value is relatively small, it is sufficient to validate our fabrication method.

 

Fig. 6 (a) Measured transmission spectrum of the compound PC microcavity, the black dot line is measurement data, the red line is fitting line with Lorentzian line-shape; (b) 3D-FDTD simulation data of transmission spectrum of the compound PC microcavity, the red line is fitting line with Lorentzian line-shape.

Download Full Size | PDF

4. Summary

In summary, we have presented a general method for making semiconductor-polymer compound NPC slab by utilizing nano-imprint lithography technique, which can help to realize perfect infiltration of thermoplastic polymer during the nano-imprint process. We have demonstrated that silicon-polystyrene compound complete NPC and NPC microcavity can be fabricated with highly uniform infiltration of polystyrene. Both the cross-sectional SEM analysis and optical transmission measurement confirm the high-quality infiltration. The transmission contrast between pass band and forbidden band of the complete photonic crystal slab is more than 15 dB and the quality factor of the compound photonic crystal microcavity is about 227, and both results are in good agreement with the simulation values. This filling method can be applied to most of semiconductors and thermo-plastic nanocomposite polymer materials in bringing together the optical property advantages of each compound material. In addition, this NIL technique is applicable to fabrication of all silicon photonic crystal structures and devices. Therefore, we expect that this technology will open up a novel route for achieving nano-photonic components with ultrafast tunable and switchable optical properties and pave the way towards building all-optical integrated devices and circuits.