As a systematic approach to demonstrate the proposed fabrication process flow and depth effect of the RF metastructure resonators, we designed the resonators using numerical solutions within a specific resonance frequency range of 4–6 GHz. We analyzed the numerical solution results regarding miniaturization, Q-factor, and resonance frequency range change as a function of their cross-sectional aspect ratios. In order to conduct resonance measurements of the resonators, we selected a microstrip ring as an antenna. After optimizing the geometrical dimensions of the microstrip ring to provide a maximum electric field in the coupling gap, our model RF metastructure resonator was placed over the gap region to couple it in the near field. Following the completion of the numerical solutions, the systematic approach's next phase involved implementing the RF metastructure's fabrication process. In addition to the two-photon lithography technique, thick-film metal deposition with electroplating and seed layer removal with dry etching were included to implement the numerically simulated structures.
We have selected indium tin oxide (ITO) coated glass as the substrate as a need for a transparent conductive seed layer (Fig. 1a). The thickness and the conductivity of ITO over the glass were approximately 100 nm and 16 Ω per square, respectively. After cleaning the substrate, we spin-coated positive photoresist AZ-4562 (MicroChem Corp.) on the substrate at 2,000 rpm for 40 s. Then the pre-baking at 110° C oven for 150 s was carried out.
We established deep trenches using the two-photon lithography 3D-printing system (Microlight 3D). The 2PP technique is the non-linear light intensity of the simultaneous absorption of two photons while creating the pattern. There was no required energy for the chemical activation of the photoresist in the focalization cone. Unlike other lithography techniques, the required intensity to initiate the photosensitive material chemical reaction existed in the focal spot. The chemical reaction, that is, functionalization of diazotnaphthoquinone (DNQ) and carboxylic acid groups (COOH), starts with the help of the laser having 532 nm wavelength at the voxel (small volume near the focal spot) of this photosensitive material. The chemically activated exposed region becomes more soluble and is then removed by a developer solution to reach the desired pattern. We can modify the trajectory of the laser beam, which determines the laser's path for the photosensitive material's chemical reaction. Modifying the path enables us to move the voxel in any direction during the patterning procedure, thereby granting us 3D-patterning capabilities. Furthermore, this adjustment also facilitates the attainment of flat sidewalls of deep trenches, which is an important requirement for implementing designs with high-aspect ratios.
As shown in Fig. 1b, we placed the prepared sample face-down to avoid diffraction from the substrate and provide steeper deep trenches. During the exposure of the photoresist, the laser focal point was moved from the top to the bottom layer through photosensitive material to increase the design quality. Laser gain (power corresponding to the used objective and photoresist), exposure time, and trajectory are critical to implement the desired 3D-designs. After adjusting all these parameters specifically for our pattern specifications (laser gain: 0.055, exposure time: 1.5 ms, filling interior horizontal/vertical: 1 µm), the printing was completed.
The following step in the process flow was the photoresist development to prepare the deep trenches using AZ-400K developer (MicroChem Corp.) mixed with deionized water in a 3:1 ratio for 5 min (Fig. 1c). This solution rate was deemed necessary due to the high-aspect-ratio structure of the sample, which required a rapid development of the exposed region without the risk of over-development ruining the flatness of trenches. The developed samples had a high-aspect-ratio, a width of 2–3 µm, and a thickness of 10–11 µm, as displayed in Figs. 2a and 2b. At the metallization stage over the deep trenches, the electroplating technique was preferred for thick-film deposition, which provides an advantage in implementing of complex 3D-structures (Fig. 1d).
The setup for metal film deposition comprises a voltage source, a copper anode, a hot plate stirrer, wires, and a copper electroplating solution. In the electroplating process, we utilized a Keithley 2400 as the voltage source, which allowed us to administer customized voltage pulses for our electroplating process. Here, the voltage source with current limiting capability was preferred to move copper ions from the copper plate (anode) to the ITO seed layer (cathode) within the high-speed bright copper electroplating solution. This type of source customization yielded superior results in scenarios involving high-aspect-ratio patterns. In addition, the bath solution was heated at 50°C to provide the required energy for ions to move around during plating, also further assisted by using a low-rate stirrer providing uniform distribution across the solution. For the current-controlled voltage source output of 100 mV with 100 µA current limitation, we found the deposition rate to be 750 nm per min. Samples with approximately 2, 4, and 8 µm metal thickness, as shown in Fig. 4, were fabricated by adjusting the electroplating duration.
After thick metal film deposition, we diced the substrate into pieces, each size 8 mm \(\:\times\:\) 8 mm in size to take experimentally accurate measurements by using a dicing saw. Since the samples had high-aspect-ratio metal structures, AZ-4562 photoresist as a protection layer was coated to protect the metal structures during the dicing operation, shown in Figs. 1e and 1f. After the removal of the photoresist (Fig. 1g), the seed layer, ITO, had to be etched to have the coupling only with the designed metal structures as shown in Fig. 1h. Inductively coupled plasma (ICP) etch was employed. We used Ar and CF4 gases with flow rates of 18 sccm and 2 sccm, respectively, while RF power was set to 500 W, DC power to 300 W, and the pressure to 20 mTorr. The etching rate was approximately 15 nm per min for ITO seed layer. Subsequently, rapid thermal annealing at 500°C for 10 min was applied to strengthen the bonds of copper, which enhanced the conductivity of the high-aspect-ratio metastructure (Fig. 1i). The length of one side of the resonator, the longest dimension, and the metal width, the shortest dimension, are approximately 4300 µm and 2 µm, respectively, indicating an enormous ratio of 2,150, as illustrated in Figs. 3a, 3b and 3c. The proposed method offers unique design capabilities when compared to structures fabricated through other techniques, such as EBL, optical lithography, and NIL.
Deep trenches are shown in Fig. 3 and Fig. 4. Here, adjustments to the electroplating process, which are the operational parameters of current-controlled voltage source, heating and stirring the solution, enhanced the uniformity in the deposition process. Heating the solution also increases the electroplating deposition rate, which means heating can lead to thicker coating at a particular time. Thus, the following optimization of time duration has to be considered together with the temperature of the solution. Also, high stirring rates can damage structures with the mixing force of the liquid since high aspect-ratio structures are fragile. The SEM images of our fabricated structures in Fig. 4 show that the resonators possess nicely flat metal sidewalls and high-aspect ratios of around 1, 2, and 4, respectively.
Our fabrication approach offers the opportunity to build RF metastructures that allow to tune resonance frequency, raise the Q-factor and reduce the footprint. Metastructure resonators in the resonance frequency range of 4–6 GHz were simulated in CST Microwave Studio to show the importance of the aspect ratio with a comparative analysis of their resonance frequency shift, Q-factor and footprint as a function of the aspect ratios (metal thickness to width ratio). Then, the designed RF metastructures having distinct aspect ratios were fabricated.
The exact value of the Q-factor can be calculated as the resonance frequency divided by the bandwidth where the reflection coefficient is 3 dB higher than it is at the resonance.35 Results show that the Q-factor of the resonator with a 2-aspect ratio is 6–7 times larger than the resonator with a 0.25-aspect ratio. Also, we examined a larger resonance frequency shift of 200 MHz varying aspect ratios of the RF metastructure resonators in both numerical simulations and experimental measurements thanks to the third-dimension effect. The resonance frequency rises as the metal thickness of the resonator grows because the product of inductance and capacitance decreases. According to the analytical approach for inductance calculation32, as metal thickness grows, the inductive effect diminishes. The capacitive impact, on the other hand, grows according to the capacitance calculation34. Because the rate of reduction in the inductor value is larger than the rate of increase in capacitance, the resonance frequency increases with increasing metal thickness. Therefore, using the third-dimension effect from the deep trenches, the resonance frequency of the metastructure designs can be fine-tuned. Here, Figs. 5a and 5b show frequency tuning and improved Q-factor of the resonators with the aspect ratio.
As a result, we have tested our process flow with the deep-trenched RF metastructures. Implementing RF metastructures also shows the importance of the cross-sectional aspect ratios based on metal thickness in terms of Q-factor and tuning resonance frequency. The result supports the idea of the effectiveness of the deep trenches used in RF metastructures.
Figure 6 shows two different resonators with the exact substrate sizes: an SRR RF metamaterial, produced using conventional LPKF PCB machine techniques and an RF metastructure created using our proposed methodology. The width of the RF metastructure was determined to be sub-10 µm with a 1.2 mm split dimension and 4.3 mm one-side length. However, the LPKF machine had a constraint on the metal width, with a minimum width of 0.6 mm. Therefore, the resonator, depicted in Fig. 6a, was fabricated with a minimum width dimension of 0.6 mm and a split dimension of 1.2 mm to demonstrate miniaturization. Besides, to adjust the resonator's resonance frequency, we increased the one-side length to 5.8 mm, as a longer length increases inductance.
The resonance frequency decreases when the metal width of the resonator is reduced from 0.6 mm to sub-10 µm while the aspect ratio remains constant. However, the Q-factor of the resonator was also decreasing in this case. Because of that, we could not obtain the desired resonance frequency. To compensate for this situation, the Q-factor was boosted with the help of high aspect ratio metal structures. Thus, increasing the aspect ratio of the resonator was required to observe resonance frequency for sub-10 µm case by increasing the metal thickness.
Analyzing metal footprints is also essential for circuit implementation in structures with a limited area to obtain a similar resonance frequency with the same range of Q-factor. Figures 6c and 6d show that RF metastructure resonator, having the third-dimension effect, acquired the same resonance frequency range of 5 GHz with a better Q-factor when the width of the metal structure was reduced from 0.6 mm to sub-10 µm as depicted in Fig. 6a and Fig. 6b. Here, although the length of one side of the SRR-LPKF is 5.8 mm, one side of the square RF metastructure is 4.3 mm. The findings reveal that RF metastructures using 3D-printing have small device footprints. This means that the resonator's footprint was reduced from 33.64 to 18.49 \(\:m{m}^{2}\), a reduction of 45%.