1. Introduction

The rapid research progress made in recent years in artificially structured sub-wavelength metallic structures, or metamaterials, has been driven by the desire to make materials that possess electromagnetic properties that cannot be found in nature. The ability to directly control the effective permittivity (ε) and permeability (µ) of these materials through top down fabrication technology has opened up many potential applications such as negative refractive index [1], optical magnetism [2], slow light [3], cloaking [4], superlensing [5], broadband polarizers [6] and sensing [7,8]. The majority of the metamaterials that have been demonstrated thus far have been fabricated using planar lithographic techniques such as electron beam lithography or UV lithography [9]. In an attempt to increase the interaction length between the impinging electromagnetic radiation and the metamaterial, research has recently moved towards extending fabrication technologies to the third dimension. Two notable examples are the multilayered fishnet structures fabricated by Valentine et al. [10] and three dimensional chiral metamaterials that have been fabricated by two photon lithography (TPL) [6].

Three-dimensional direct laser writing utilizing TPL is particularly appealing as it is the only technique that allows for arbitrary three dimensional control over the fabrication process [11]. The technique utilizes nonlinear optical absorption of a tightly focused pulsed laser beam to selectively crosslink a small volume inside a polymer. By moving the focal point relative to the sample in all three dimensions, arbitrary three-dimensional nanostructures can be fabricated. Features down to 23 nm have been reported by Tan et al. [12] and aspect ratios as high as 50 have been shown in SU-8 [13]. Due to its high resolution and the high level of control, TPL has been recently applied to many research areas, such as photonics and plasmonics [14], biomedical microfluidics [15] and MEMS devices [16]. Other direct write fabrication technologies such as proton beam writing (PBW) have also been recently applied to the fabrication of metamaterials [17,18]. Although high aspect ratios [19] and nanostructures down to 20 nm have been achieved [20] using this technique, three-dimensional control over the structure design is limited to a few layers.

One limitation of the TPL technique is that it can typically only be used to pattern polymers like SU-8 and Ormocer. A metallization step is needed to convert the patterns into metallic structures for metamaterials applications. Rill et al. [21] utilized a chemical vapor deposition (CVD) process for transferring the SU-8 structures to metal. The process however required the deposition of a SiO₂ layer to protect the SU-8 template from the high temperatures necessary for deposition. To achieve conformal coating on 3D structures, a more suitable approach would be electroless plating. It involves metallization through a chemical reduction process. Ag is widely adopted as the metal of choice, especially for applications in metamaterials and plasmonics, because of its high conductivity and low absorption coefficient [22]. Due to the poor adhesion between metal particles and the polymer surface, additional processes are needed for surface functionalization. In addition, selective coating is important for metamaterials applications in order to prevent the substrate from an electrical short between the structures. Radke et al. [23] have used TPL and electroless Ag plating to fabricate SU-8 chiral metamaterial structures. However, due to the lack of plating selectivity, the structures had to be fabricated on post and detached from the substrate prior to optical measurements.

Therefore various pretreatment techniques have been employed to achieve coating selectivity. Formanek et al. [24] demonstrated a number of steps needed to achieve selective electroless coating of metal on polymer instead of glass. First, it requires the production of a hydrophobic coating on glass supporting the polymer structures. Also a crucial part of the work involves the use of chemically modified photopolymerizable resin for TPP fabrication. The sample is then submerged into a tin chloride solution for activation of the polymer surface for metal reduction. Wen Dai et al. [25] applied a large UV dose to rearrange the epoxy structures of the cross-linked SU-8 surface in order to make it more adhesive to Ag. However this process is directional and cannot achieve conformal coating of 3D structures.

In this work, we demonstrate a direct approach for selective conformal coating of Ag on three-dimensional SU-8 microstructures that have been fabricated on a Si substrate. After fabricating the structures, the SU-8 surface is modified using a Radio Frequency (RF) plasma prior to electroless Ag coating. The plasma alters the SU-8 surface so that Ag nanoparticles can preferentially nucleate on the surface while leaving the Si substrate unaltered. No resin modification is required and this process can be performed on SU-8 resist that is widely available and used for two-photon lithography. The simplicity and effectiveness of this process makes it more appealing for 3D metamaterials applications.

For our experiment a 25 μm layer of SU-8 is spin-coated on a clean Si substrate. 3D chiral structures are then fabricated on the resist using our in-house TPL fabrication system. The femtosecond laser pulses used in TPL are generated by a mode-locked Ti:Sapphire oscillator (Coherent, Mira 900D) which is pumped by a 10 W 532 nm laser (Coherent,Verdi V10), with a centre wavelength of 800 nm and pulse width of approximately 150 fs. A computerized sample platform consisting of several precision stages allow the user to move in all three dimensions with nanometer precision. The chiral structures were fabricated by moving the stages in three dimensions according to a predefined path. After irradiation, the sample was post-baked at 95°C for 10 minutes before a final development and iso-propyl alcohol (IPA) rinse. After TPL, the sample was subjected to an RF plasma treatment using a plasma cleaner (Harrick Plasma PDC-32G). Since the plasma comes from the coils surrounding the sample holder, all surfaces receive the same amount of plasma dose. For electroless Ag plating, we utilized the standard Tollens’ reagent method. The plating process is conducted in a 50 mL silver nitrite (AgNO₃) solution with a concentration of 0.05 mol/L kept at a temperature of 35°C to 50°C. A glucose solution (1 mL) with a concentration of 0.025mg/mL is added into the solution to reduce the Ag ions to metal particles. After 3 minutes of reaction, the sample is taken out and cleaned.

Atomic force microscopy images were used to measure the surface roughness of the coated SU-8 sample. In order to find the coated surface more easily, a Si substrate was first spin-coated with SU-8 and cut into pieces. Then the SU-8 samples were UV exposed, plasma pretreated and electroless Ag plated. The AFM scan area was set to 5.0 × 5.0μm² and the root mean squared (RMS) surface roughness value was measured.

Fourier Transform Infrared Spectroscopy was carried out using a Bruker Vertex 80v vacuum FTIR system. The terahertz beam produced from a mercury arc lamp was discriminated into transverse electric (TE) and transverse magnetic (TM) polarization using a polyethylene FIR polarizer placed before the sample. The transmitted signals were then detected with a liquid helium cooled Si bolometer (Infrared Laboratories) at a resolution of 0.2cm⁻¹.

2. Results and discussion

One indicator of how well the surface is coated by the electroless plating process is the coating coverage. The coating coverage, defined as the ratio of coated area to the total surface area, can be determined from the Scanning Electron Microscope (SEM) image. Figure 1 shows a plot of Ag coverage as a function of applied plasma dose for a Si substrate and cross-linked SU-8 polymer. If no plasma is applied, there is almost no Ag coverage on both the Si and SU-8 surfaces after electroless Ag plating for 3 minutes. As the RF plasma dose is increased, we find that the Si and SU-8 surfaces behave differently after the treatment. By increasing the plasma dose, we see that there is hardly any silver on the Si surface. On the other hand, a significant increase in the silver coverage of SU-8 can be observed. At a dose of 420 J, 42% of the SU-8 surface has been covered by silver. As the dose increased further to 2160 J, the SU-8 surface becomes completely covered, while leaving the Si surface relatively free from silver particles. This can be seen clearly from the SEM images in the inset of Fig. 1. We measured the resistivity of the coated SU-8 surface using the four point probe method and found that the Ag coated layer of 100 nm had a value of 9 × 10⁻⁸ Ω.m. This is of the same order as the nominal value for bulk Ag which is 1.6 × 10⁻⁸ Ω.m. Ellipsometry measurements were also performed in order to compare the optical properties of the silver electroless coated SU-8 films under optimized conditions, with films of similar thickness prepared using electron beam evaporation. Both films exhibited very similar optical properties.

 

Fig. 1 Coating coverage of Si and SU-8 surfaces as a function of plasma treatment dose. Insets: SEM images of the Si and SU-8 Ag coated surface using a 2160 J plasma treatment.

Download Full Size | PDF

To understand the modification of the Si and SU-8 surfaces under plasma treatment and how they interact with Ag ions we look to results obtained by previous studies. X-ray Photoelectron Spectroscopy measurements performed by Walther et al. [26] have shown that an additional peak centered at 289.2 eV appears for the plasma treated SU-8. This peak can be attributed to carboxylic (COOH) and aldehyde (CHO) groups generated by oxidation. Gerenser et al. [27] showed in an earlier study that Ag ions can interact with C = O bonds in the carboxylic and aldehyde groups to form a Ag-O-C bond. It is through this mechanism that the SU-8 surface becomes a preferential catalyst site for the reduction of Ag ions (See Figs. 2(a) –2(c)). On the Si surface, the plasma treatment produces hydroxyl (-OH) groups that can account for the hydrophilicity of Si, however this does not increase the adhesion of Ag to the surface [28]. Figures 2(d)–2(g) show conformal coating of three-dimensional chiral metamaterials fabricated using TPL before and after our selective electroless plating process. It can also be seen that the substrates are relatively free from Ag particles.

 

Fig. 2 Mechanism and results of the selective electroless Ag coating method. a) Plasma treatment induces C = O bond on SU-8 surface; b) Ag ions in the solution attach to SU-8 surface and interact with C = O to form Ag-O-C bond; c) The SU-8 surface becomes a preferential site for the reduction of Ag when glucose is added to the solution; d) and e) A three dimensional chiral structure fabricated using TPL before and after being coated with Ag; f) and g) A three dimensional helical structure before and after being coated with Ag.

Download Full Size | PDF

Surface roughness is an important parameter to consider for metamaterials applications due to its contribution to scattering loss. Optimization of the plating parameters needs to be carried out in order to minimize the surface roughness. One of the most important factors affecting the roughness is the concentration of glucose since it is responsible for the reduction of Ag ions into particles and the rate at which Ag nanoparticles are generated. By keeping all other parameters constant, we have investigated the surface roughness for three glucose concentrations (0.0125mg/mL, 0.025mg/mL and 0.0375mg/mL) and at three different plating times (60s, 90s and 120s) for a constant temperature of 45°C. AFM results plotted in Fig. 3(a) show that the surface roughness increases with increasing glucose concentration, and plating time.

 

Fig. 3 (a) RMS surface roughness measured as a function of glucose concentration; (b) RMSsurface roughness measured as a function of coating temperature, for three plating times; (c) SEM image of a coated surface under un-optimized conditions; (d) SEM image of a coated surface under optimized conditions.

Download Full Size | PDF

In order to study the effect of reaction speed on surface roughness, we have also varied the temperature of the plating process. Using the optimized glucose concentration of 0.0125 mg/mL and plating time of 60 s obtained from previous measurements, we show in Fig. 3(b) that the results favor slower reaction rate and lower temperature produces smoother surface. It is found that the lowest surface roughness (9.1 nm) can be obtained at a plating temperature of 35 °C with a glucose concentration of 0.0125 mg/mL for 60 s. When the reaction rate is too high, Ag forms islands, causing surface roughness and uneven coating. It was found that for a glucose concentration below 0.0125mg/mL and a temperature below 35 °C, the production rate of Ag nanoparticles is too slow for significant coating to occur.

To demonstrate the effectiveness of the coating technique for metamaterials applications, an array of THz double split ring resonators (SRRs) was fabricated. THz metamaterials have been extensively studied in recent years [29–32] and the THz metallic SRR structures exhibit LC resonance for incident electromagnetic waves with polarization parallel to their split gaps. To verify the performance of the coated SRRs, the structures were first fabricated in a 2 µm SU-8 layer spin-coated on a Si substrate using UV lithography, followed by plasma pretreatment and then coated with a 100 nm layer of Ag. Figure 4(a) shows an SEM image of the Ag coated SU-8 SRR array, with minimal Ag deposition on the Si substrate. The FTIR spectrum, shown in Fig. 4(b), reveals the presence of an LC resonance at approximately 0.64 THz for light with the electric field polarized parallel to the SRR gap. The resonance disappears when the electric field polarization is rotated by 90°. At the resonance frequency, the skin depth of the incident light is less than the coated Ag thickness therefore the interaction between the incident THz wave and the underlying SU-8 dielectric is minimal. Simulations (not shown) reveal that at these frequencies, the difference between the resonant frequency of an SRR made from bulk metal, and one made from an SU-8 coated metal is minimal.

 

Fig. 4 (a) An array of 2D double split ring resonators with height 5 µm fabricated in SU-8 on Si and coated with Ag using selective electroless plating. (b) Transmission spectra for the electric field parallel and perpendicular to the SRR gap showing an LC resonance dip at 0.64 THz.

Download Full Size | PDF

3. Conclusion

In conclusion, we have demonstrated a simple and effective electroless Ag plating method for conformal uniform coating of arbitrary three-dimensional SU-8 structures. High coating selectivity and good conductivity has been achieved on SU-8 over Si by a plasma pretreatment of the surface prior to plating, a feature that is crucial for metamaterials applications. An RMS surface roughness of 9 nm was obtained under optimized conditions. This compares favorably with Ag surfaces that are produced using electron beam evaporation (Typically 4-8 nm RMS depending on the substrate [33]). To further improve the surface quality we are currently investigating the use of physical surface smoothing techniques [34]. An SRR array fabricated using this technique shows optical properties similar to bulk electroplated or evaporated metals. Combined with TPL laser fabrication, this process opens up numerous possibilities for true three dimensional metamaterials structures.