1. Introduction

There have been increasing demands on developing effective biological sensors for various applications such as classification of specimens, verification of toxicity, and discovery of cell growth mechanism [1–5]. Yeast has been an ideal sensing element for developing and optimizing novel microbial sensors, because it is robust, grow rapidly on a broad range of substrates, and provides information directly relevant to other eukaryotes cells [6]. Terahertz (THz) spectroscopy is a powerful technique for non-contact, non-destructive, and label-free examination of biological and chemical substances. However, THz spectroscopy has not gained extensive consideration as a detection tool for microorganisms because they are transparent in the THz frequency range and their typical size is on the order of or less than λ/100, resulting in low scattering cross-section. Recently, we were able to visualize the microorganisms such as yeast, molds, and bacteria in the THz frequency range, with the help of metamaterial sensors [7]. It was possible because the resonant frequency of metamaterials is highly sensitive to changes in the dielectric constant of the gap area [8] and due to size compatibility of the microorganisms to the gap size. In particular, yeast cell has a high THz dielectric constant and hence is ideal to test and optimize the novel microbial sensors operating in THz frequency range [7].

On the other hand, THz slot antennas consist of periodically arranged sub-wavelength rectangular holes and show exclusive electromagnetic properties such as shape resonance [9, 10]. The resonant transmission in slot antenna structures is associated with extremely localized and enhanced fields [11–18] that enable sensitive sensing of conducting or dielectric materials [19]. Here, the resonance of the slot antenna is determined mainly by its length, whereas the field localization and enhancement can be controlled by its width. Therefore, to achieve the optimal sensitivity without significant change in resonant frequency, it is possible to adjust the width of the slot antenna such that it is comparable to the size of the target materials. In addition, unlike metamaterial sensors, THz slot antennas allow the transmission only through slot areas where the biological substances are located, resulting in reduced background signals. Thus far, THz slot antenna arrays have not been demonstrated as efficient microbial detectors. Moreover, the effects of geometrical factors, such as the width of the antenna and the substrate index, on the sensitivity have not been studied in plasmonic and metamaterial sensing. The effects of the substrate index on the sensitivity are of particular interest since the resonant frequency is strongly linked to the substrate index [20, 21].

In this paper, we present THz time-domain spectroscopy (THz-TDS) studies of slot antenna sensors for highly sensitive detection of microorganisms such as yeast cells. We observed a resonant frequency shift in the shape-resonance upon deposition of yeast cells. In particular, THz slot antenna sensors were fabricated on various substrates with different dielectric constants, in which we studied the substrate effects on the sensitivity. In addition, the frequency shift was investigated as a function of the number density, thickness of yeast film, and geometrical parameters of the antenna. The experimental results were confirmed by finite-difference time-domain (FDTD) modeling of the same structure.

2. Experimental results and discussions

The real-time THz transmission amplitudes of the slot antenna with yeast cells were measured using a conventional THz-TDS setup [22]. A femtosecond laser with λ = 800 nm is incident on the photoconductive antenna, which emits a linearly polarized THz pulse. The pulse is then focused on the slot antenna array with ~1 mm² spot area under ambient conditions. Time traces of the transmitted THz electric field, both in amplitude and phase, were measured by varying the time delay between the probe beam and the THz pulse. The THz spectrum is obtained by applying a fast Fourier transform (FFT) to the time trace and normalized with respect to the reference.

The yeast samples were grown by a streaking on medium method, followed by incubation at 37 °C for 2 days. The culture medium was yeast peptone dextrose. Yeast was obtained from the Korean Agricultural Culture Collection (KACC). THz slot antenna patterns were prepared by using a conventional electron beam lithography method on undoped Si (n-type, resistivity > 6000 Ω·cm, and thickness of 550 μm) and quartz (thickness of 2000 μm) substrates, followed by metal deposition of Cr/Au (2 nm / 98 nm) using an electron beam evaporator. The 10 × 10 slot antenna arrays consist of slot antenna patterns with length (l) of 100 μm and periodicity of 200 μm, while we varied the width (w) at 2, 5, 10 μm. THz transmission spectra were obtained by using THz-TDS techniques with an acquisition time of 5 s for each spectrum [22].

Figure 1(a) shows a schematic of the THz slot antenna sensing of yeast cells. We measured the change in the spectra of the THz radiation transmitted through the slot antenna after the deposition of yeast cells. Without the cells, the resonance of the transmission is determined mainly by the length of the slot antenna while it is modified by the substrate refractive index (n_sub). By introducting an effective refractive index n_eff, the resonant frequency (f₀) can be expressed as f₀ = c/(2n_effL), whereas n_eff can be determined as a linear combination of the substrate and air refractive indices, as reported previously [20, 21]. Dielectric materials located in the slot area induce a change in the effective dielectric constant of the antenna. As a result, a resonant frequency shift (Δf) occurs in the THz transmission when the yeast cells are present in the slot area of the antenna.

 

Fig. 1 (a) Schematic of the THz slot antenna sensing of yeast cells. (b) SEM image of the yeast cells deposited on the slot antenna device with antenna width (w) of 2 μm, and length (l) of 100 μm. (c) Normalized THz transmission amplitudes for the THz slot antenna on Si substrate, with (red solid line) and without (black solid line) yeast cells.

Download Full Size | PDF

A scanning electron microscopy (SEM) image in Fig. 1(b) shows the typical slot antenna device with the width of 2 μm, when it is covered with yeast cells (dark circles). Yeast cells were deposited on the devices by rubbing a yeast-coated swab on them, Figure 1(c) shows normalized THz transmission amplitudes of THz slot antenna devices with w = 2 μm, fabricated on Si substrates. Compared with the yeast-free case (black solid line), the device with yeast cells on the slot antenna arrays (red solid line) shows a noticeable red shift of the resonant frequency by about 9 GHz when the average number of yeast located in the slot area (N_av) is 50. N_av was determined by counting them for the entire set of elements (6 × 6) positioned in the spot area of the THz fields (~1 mm²).

To optimize the device sensitivity, it is important to investigate the resonance shifts as a function of the substrate dielectric constants and geometrical parameters of the antenna. Because the detection of microorganisms is based on dielectric sensing, changes in the dielectric substrates will strongly influence the sensitivity. Figure 2 shows the shift of the resonant frequency as a function of N_av for two different types of substrates. First, Fig. 2(a) shows a series of transmission spectra for the device fabricated on a Si substrate. The red shift of the resonant frequency (f₀ = 0.57 THz) increases as N_av increases. Similar experimental results were found with the slot antenna arrays fabricated on a quartz substrate (f₀ = 0.93 THz), as shown in Fig. 2(b). Surprisingly, we found that the resonance shift is more pronounced and approximately 3-fold higher than that of the Si substrate device. Figure 2(c) shows a plot of Δf as a function of N_av for Si (black boxes) and quartz (red triangles) substrates. For both types of substrates, the magnitude of Δf increases gradually with N_av and, for a given N_av, Δf is 2.83 times (1.74 times, in terms of Δf / f₀) higher for the quartz substrate. This result represents that the sensitivity of the quartz substrate slot antenna is higher than that of the Si substrate case. This is because the relative change in the effective dielectric constant with the presence of the yeasts is higher for the substrates with lower permittivity.

 

Fig. 2 Normalized transmission amplitudes for the THz slot antenna array on (a) Si substrate and (b) quartz substrate at different yeast number densities, N_av. (c) Resonant frequency shift as a function of N_av for Si (black) and quartz (red) substrates.

Download Full Size | PDF

The shifts in resonant frequency can be explained by considering the relationship Δf / f₀∝N(ε_f – ε_air)/ε_eff, where $N$is the number of yeast cells; ε_f is the dielectric constant of the individual yeast cell; ε_air is the dielectric constant of the air; ε_eff ( = n_eff ²) is the effective dielectric constant without yeast cells in the slot area [7]. Evidently, Δf / f₀ is inversely proportional to ε_eff, and thus the sensitivity can be improved by using substrates with lower dielectric constants. As mentioned before, n_eff can be represented by a linear combination of the substrate and air refractive indices. As the contribution from the substrate index is about twice that of the air index, we estimated that ε_eff is equal to 7.56 and 2.96 for the Si (n_sub = 3.38) and quartz (n_sub = 1.93) substrates, respectively [21]. However, this result implies that Δf / f₀ is about 2.55 times higher in the quartz substrate case than in the Si case, while a value of 1.74 has been obtained from our experimental results in Fig. 2. This discrepancy is likely to originate from the fact that the size of the yeast (~2 μm) is much larger than the thickness of the metal film (100 nm), and thus the cells are located far into the air side. This configuration will obscure the substrate effects as described by the simple equation relating Δf to ε_eff as shown above. However, as it will be shown later, the results in Fig. 2 are in good agreement with the simulations, where we take into account the shape of the microorganisms.

Subsequently, we studied the effects of the geometrical factors and the vertical range of the effective sensing volume by using thin films consisting of closely packed yeast cells. THz transmission experiments were performed on yeast films deposited from a dense yeast solution, followed by a drying process in a vacuum oven for 6 min. Figure 3(a) shows a schematic drawing of the experimental design used to investigate the dependence of the peak shift on film thickness (t) and slot width. To demonstrate the effect of the slot width on the sensitivity, which is associated with the field enhancement factor [19, 23], we used slot antenna arrays with three different widths (2, 5 and 10 μm) and with the same antenna length (l = 100 μm). Figure 3(b) shows the resonant frequency shift as a function of the slot antenna width for both types of substrates. The thickness of the deposited yeast films was the same (t = 4 μm). In both cases, the frequency shift decreases significantly as the width increases. These results confirm that slot antenna patterns with smaller width are more favorable in microbial sensing although it is restricted by the size of the individual microorganisms. It is also clear that development of slot antennas with nanoscale width will be necessary for sensing submicron-sized target materials such as virus.

 

Fig. 3 (a) Schematic of the experiment performed to investigate the effects of yeast film thickness (t) and slot antenna width on the peak shift. (b) Resonant frequency shift as a function of w for Si (black) and quartz (red) substrates with t = 4 μm. (c) Resonant frequency shift as a function of tfor Si (black) and quartz (red) substrates.

Download Full Size | PDF

Figure 3(c) shows Δf as a function of the film thickness, in the range 0 – 16 μm. For both substrate types, Δf increases with the film thickness, reaching saturation at a characteristic thickness value of t_c = 3.5 μm (w = 2 μm), where t_c is defined from the relation Δf ∝1–exp(t/t_c). This result represents a measurement of the effective detection range with respect to the substrate. Although the peak shift at the fixed film thickness varies with different sample conditions owing to the different wetting conditions, both types of substrates demonstrate similar characteristic lengths. The knowledge of the thickness-dependent sensitivity is critical in finding the optimal design of the microbial sensors, especially in aqueous environments. As mentioned before, THz plasmonic sensors are highly sensitive to analytes near the surface, and therefore they will serve as an effective platform for sensing microorganisms in aqueous environments because they enable the use of a very thin water layer.

To study the electromagnetic response of the THz slot antenna, the structure was modeled by using Lumerical FDTD simulation. We used linearly polarized plane wave as a source and applied periodic boundary condition to simulate an array of slot antenna structure. As shown in Fig. 4(a), we used the slot antenna patterns with the same geometric parameters utilized in the experiments with the metal films considered as a perfect electric conductor. To replicate the yeast cells, we introduced a series of spheres of diameter 2 μm, as shown in the inset of Fig. 4(a). The dielectric constant of the dielectric spheres was chosen to be ε_f = 8 [7].

 

Fig. 4 (a) FDTD simulation results of transmission spectra through slot antenna array on quartz substrate with (red) and without (black) yeast (N_av = 50). (b) Δf as a function of Nav, for Si (black) and quartz (red) substrates. (c) Δf as a function of yeast film thickness for Si (black) and quartz (red) substrates.

Download Full Size | PDF

As shown in the normalized spectra in Fig. 4(a) for N_av = 50, in the case of the quartz substrate, the resonant frequency of the THz slot antenna exhibits a clear red-shift when the dielectric spheres are present. To further confirm the experimental results, we performed a series of simulations to investigate the resonance shift as a function of N_av for different types of substrates, as shown in Fig. 4(b) for Si (squares) and quartz (triangles). The detection of yeast cells using slot antenna patterns on quartz substrate was more sensitive than the detection using Si-based devices by about 2.93 times. This result is very close to our experimental value of 2.83. In other words, by taking the shape of the yeast into account, the relative difference in sensitivity between the two substrates has been successfully reproduced. Although all geometric parameters are identical, the overall resonant frequency shift obtained from the simulations is generally smaller than the one measured in the experiments, for both types of substrate. The discrepancy is possibly caused by simplifications in the shape of the yeast cells. Finally, Fig. 4(c) shows the dependence of the resonance shift on the film thickness. Instead of using a collection of dielectric spheres, we considered simple dielectric films with a net dielectric constant ε_f,film = 3.85 (f₀ = 0.57 THz) and ε_f,film = 3.62 (f₀ = 0.93 THz) for Si and quartz substrates, respectively [7]. Here, the estimated characteristic length of ~2.8 μm, for both types of the substrates, is in reasonable agreement with our experimental results.

3. Conclusions

THz plasmonic structures such as slot antenna arrays can be used as effective sensors to detect low-density microorganisms such as yeasts. A clear shift in the shape resonant frequency, due to the change in the effective dielectric constant at the slot area, is observed following the deposition of yeasts. We found that THz plasmonic sensors based on low-dielectric constant substrates (e.g. quartz) demonstrate higher sensitivity than sensors based on high-dielectric constant substrates (e.g. silicon). For yeast films of the same density, the sensitivity is higher for smaller slot widths, confirming the field enhancement effect. The measured effective vertical range confirms that the detection volume is highly localized near the surface. THz slot antenna sensing is a background-free method because the transmission occurs only through the apertures in which the biological substances are located. In addition, the sensor is highly sensitive to the size of the target materials, and its sensitivity can be easily controlled by tailoring the width of the antenna, without changes in resonance. Our work will contribute in the development of effective sensors and their optimization, for rapid, accurate and on-site detection of hazardous microorganisms in diverse environments.