1. Introduction

Terahertz (THz) electromagnetic waves, which possess frequencies ranging from a few hundred gigahertz to a few tens of THz, have excellent potential to be applied for molecule sensing because numerous molecules have unique spectral fingerprints, i.e., absorption peaks, in this frequency range [1]. The molecules can be detected by comparing the transmittance or reflectance spectra of the THz waves before and after they interact with the molecules. Despite the existence of many promising applications based on this technique, such as the sensing or monitoring of chemical or biological molecules, there exists a limitation in that the changes in the spectra caused by interaction with the molecules are very weak. To overcome this limitation, metamaterials, i.e., artificial metallic or dielectric structures that can manipulate the propagation of the wave such for invisibility cloaking [2–4] or high-resolution imaging [5–7], can be utilized to enhance the variations in the spectra. For instance, metamaterials based on slot antennas can increase the absorption cross sections of molecules by magnifying the electromagnetic field inside the gap of the slot [8–10]. In addition, molecules can be detected by measuring the shift in the resonant frequencies of metamaterials using spoof surface plasmon polaritons [11], Fano resonance [12], or Fermi level in graphene [13].

Even though the performance of THz sensors assisted by metamaterials has been verified successfully, there still remains an inherent constraint in that the THz spectrum has to be measured and analyzed via THz time-domain spectroscopy (TDS). Owing to their bulkiness, the current THz-TDS systems could hinder the diagnosis of molecules in mobile platforms. To realize a compact system, such as a kit for the in-situ sensing of the avian influenza virus using the near-infrared spectrum [14], the THz sensor has to be simplified. A pyroelectric sensor [15] may provide a solution to simplify the system, however, there exists a limitation that the configuration of metamaterial absorber separated from the pyroelectric film may degrade the sensitivity of the sensor. Although a sensor utilizing the superconducting transition edge [16] can be an alternative with respect to sensitivity, an optical cryostat adopted for superconducting effect at extremely low temperatures may increase the complexity of the system. Not only to improve the sensitivity but also to simplify the system, detectors or imaging apparatus that utilizing the thermal energy converted from THz waves have been reported successfully based on a thermally sensitive microcantilever [17], a THz-to-IR converting metasurface [18], an all-dielectric metasurface absorber [19], and a multispectral metamaterial absorber [20].

Herein, a metamaterial bolometer is proposed that can detect variations in the power of THz radiations by measuring the electrical resistance of the bolometric layer at room temperature. To effectively convert the power absorbed from the irradiated THz waves into the variation of the electrical resistance, a thermosensitive composite, namely, polydimethylsiloxane (PDMS) mixed with carbon black (CB) [21], is used as the substrate of the proposed metamaterial absorber. The electrical resistance of the bolometer increases in accordance with the increase in temperature due to the absorption of the THz waves. If molecules attenuate the THz waves irradiating the bolometer, they can be detected via the variation of the electrical resistance of the bolometric layer. Experiments conducted using a fabricated bolometer chip show that the detection of THz waves based on the bolometric method is possible even using a low-power THz source. The resistance increases with the incidence of the THz waves and reaches a steady-state value that is sufficiently larger than the change in resistance induced by the background thermal noise. The proposed THz bolometer may serve as a novel method to realize a mobile sensor that could detect molecules having fingerprints in the THz gap.

2. Design and simulation of terahertz metamaterial bolometer

Figure 1(a) shows the configuration of the proposed metamaterial bolometer. First, a 500 nm-thick silicon dioxide (${\textrm {SiO}}_2$) layer was deposited on the bottom of 250 µm-thick silicon (Si). A two-dimensional (2D) gold metamaterial structure was patterned on the ${\textrm {SiO}}_2$ layer via photolithography [22], and the Si substrate was chemically etched. The detailed structure of the metamaterial will be described later in this section. The ${\textrm {SiO}}_2$ layer on the Si substrate is used as an etch-stop layer [23], as shown in Fig. 1(a). Furthermore, after sputtering a 100-nm-thick ${\textrm {SiO}}_2$ layer on the gold pattern, a bolometric layer is coated on the ${\textrm {SiO}}_2$ layer. The sputtered ${\textrm {SiO}}_2$ layer insulates the metal pattern from the electric current in the bolometric layer. The bolometric layer consists of $45\textrm {wt}\%$ PDMS and $55 \textrm {wt}\%$ CB microparticles, and demonstrates a high temperature coefficient of resistance (TCR) ranging from 1.88$\%$/K to 3.11$\%$/K at room temperature (22.2-23.8$^\circ$C), as shown in Fig. 1(b). To obtain the TCR value, the resistance was differentiated with respect to temperature and divided by the resistance. The thickness of the CB-PDMS layer should be minimized such that it is within the range that guarantees sufficient THz waves absorption for highly sensitive detection of the temperature variation. In this setup, the CB particle size ranges from 2 to 10 µm, and the minimum thickness of the CB-PDMS layer guaranteeing uniform coating via an applicator was 50.8 µm. The performance of the layer with regard to THz waves absorption will be discussed later. On the bolometric layer, a pair of gold electrodes was patterned via the sputtering process with a shadow mask and electrically connected to a digital multimeter to measure the resistance of the layer.

 

Fig. 1. Schematic and physical properties of the proposed metamaterial bolometer. (a) Cross-sectional schematic, (b) electrical resistance with respect to temperature, and (c) refractive index of CB-PDMS modeled by effective medium approximation.

Download Full Size | PDF

To design the metamaterial patterned on the bolometric layer, which can maximize the absorption of the THz waves, the electromagnetic properties of the CB-PDMS layer should be known. For this purpose, a well-known effective medium approximation formula, namely, the Maxwell-Garnett formula [24–26], was used. In this study, the real part of the effective permittivity of the mixture was calculated based on the mixing formula using the permittivity of PDMS, which includes a $10\%$ curing agent [27], and amorphous carbon [28]. In contrast, the imaginary part of the permittivity was calculated using the effective conductivity measured via the four-point probe method, which can measure the sheet resistance of the thin film. The calculation of the imaginary part using the mixing formula was not appropriate for the CB-PDMS mixture because the size of the CB particle is quite large compared with the skin depth of it [25]. The average and the standard deviation of the sheet resistance of the CB-PDMS layer were measured to be 584.7 $\Omega$/sq and 53.18 $\Omega$/sq, respectively. By using the average of the sheet resistance, the effective conductivity $\sigma _{r}$ of the mixture was calculated as $\sigma _{r}=1/(R_s\cdot d)$, where $d$ is the thickness of the layer (50.8 µm) and $R_s$ is the average sheet resistance. Finally, the effective complex permittivity was calculated as $\epsilon _{r}= \epsilon '-i\sigma _{r}/({\omega }\cdot \epsilon _0)$, where $\epsilon '$ is the real part of the effective permittivity calculated by the mixing formula, $\epsilon _0$ is the permittivity of free space, and $i$ indicates $\sqrt {-1}$ [29]. Figure 1(c) shows the refractive indices of CB-PDMS modeled by the effective medium approximation in the range from 0.5 to 1.5 THz.

The unit cell structure of the metamaterial absorber was designed under consideration of the electromagnetic properties of the CB-PDMS layer, and its dimensions are shown in Fig. 2(a). All the boundaries of Fig. 2(a) were set to satisfy the periodic condition for the full-wave simulation using the commercial software Comsol Multiphysics. The magnitude of the incident electric field $|E_y|$ was 1 V/m. In contrast with conventional metamaterial absorbers [22,30–32], the proposed metamaterial absorber does not have a metal reflector because a gap is needed between the electrodes, as shown in Fig. 1(a), to measure the resistance of the CB-PDMS layer. This configuration may limit the performance of the layer with regard to the absorption of the THz waves. Nevertheless, it is possible to minimize the heat capacity of the absorber and the thermal resistance between the part that absorbs the THz waves, i.e., the metamaterial pattern, and the bolometric layer by omitting the metal reflector. As the heat capacity and thermal resistance decrease when the number of layers is minimized, the temperature change with respect to THz waves absorption increases, thereby resulting in the increased sensitivity of the bolometer.

 

Fig. 2. Schematic and full-wave simulation results of the proposed metamaterial absorber optimized to a target frequency of 1 THz. (a) Dimensions of the microscale gold metamaterial pattern. (b) Surface currents on the gold pattern induced by the time harmonic THz wave polarized along the y-axis. (c) Electric fields inside gaps of the gold pattern. (d) Reflectance (R), transmittance (T), and absorption (A) of the proposed metamaterial absorber.

Download Full Size | PDF

A pair of half rectangular rings connected by a gold strip was designed to achieve THz wave-induced resonance of the electric current. The series circuit consists of the self-inductive gold pattern, and the capacitance of the gap between the half rectangular rings supports the resonant surface current when the electric field that is polarized along the $y$-axis is incident on it, as shown in Fig. 2(b). The resonance enhances the absorption of THz waves by increasing the ohmic loss. In addition, rectangular gold strips surrounding the resonator were proposed to boost the absorption. The strips aligned along the $y$-axis located at the left and right sides of the unit cell support the resonating electric current based on the same mechanism with the resonator at the center. The gaps at the four corners of the unit cell add capacitances to the self-inductances of the strips and therefore form series resonant circuits. Moreover, resonant electric fields are formed between strips aligned along the $x$-axis and the resonator at the center, as shown in Fig. 2(c). The resonant electric fields increase the dielectric loss inside the lossy bolometric layer.

The full-wave simulation results of the reflectance, transmittance, and absorption of the proposed metamaterial absorber are shown in Fig. 2(d). The absorption is maximized to 0.446 at 1 THz, which is 4.6 times greater than that of the bolometric layer without the gold-patterned metamaterial (0.097). The target frequency was set to 1 THz because most of the power of the THz waves used in this study is concentrated around 1 THz, as will be shown in the next section. The reflectance is minimized to 0.368 at the same frequency. Although there remains a non-negligible transmittance of 0.186 at 1 THz due to the absence of the backside metal reflector, the resonant response that can be found in the absorption curve enables the highly sensitive detection of the variation in the THz power around 1 THz.

3. Experimental verifications

To verify the performance of the proposed metamaterial bolometer experimentally, the THz waves were irradiated on the aperture in the front side of the bolometer chip shown in Fig. 1(a). The configuration of the experiment and the fabricated metamaterial pattern observed through the aperture are shown in Fig. 3(a). To irradiate the THz pulse, a common low-power THz time-domain spectroscopy (TDS) system driven with a femtosecond Ti:sapphire pulsed laser was exploited of which the center wavelength, the pulse width, and the repetition rate are 800 nm, 100 fs, and 80 MHz, respectively. A photoconductive antenna and an electro-optic sampling with a ZnTe crystal were employed for emission and detection of THz waves, respectively [10]. To focus the emitted THz waves on the metamaterial bolometer and receive all of the transmitted THz waves via the detector, a pair of parabolic mirrors and polymethylpentene (TPX) lenses were utilized. The peak-to-peak amplitude of the THz electric field intensity ($E_{pp}$) was 20 V/cm, which was estimated via the maximum modulation of the photodiodes signals [33] and material properties of the ZnTe detection crystal [34]. By comparing $E_{pp}$ with that of the THz pulse generated by the ${\textrm {LiNbO}}_3$ crystals [33], it can be conformed that the power of the THz pulse used in this study was relatively low.

 

Fig. 3. Fabricated metamaterial bolometer chip and measurement results. (a) Setup for measurement. Right panel: metamaterial pattern observed by an optical microscope. (b) CB-PDMS layer covered with tapered electrodes. Right panel: gap between the tapered electrodes observed by an optical microscope. (c) Spectrum of the THz waves irradiated on the metamaterial absorber. (d) Measured and calculated reflectance ($R$) and transmittance ($T$). Measured $R$: circle with solid line, measured $T$: cross with solid line, simulated $R$ with scaled $R_s$: upward triangle with dotted line, simulated $T$ with scaled $R_s$: square with dotted line, simulated $A$ with scaled $R_s$: diamond with dotted line.

Download Full Size | PDF

Because the THz beam is irradiated on the metamaterial absorber through an aperture with an area of 1 $\times$ 1 $\textrm {mm}^2$, the heat is primarily generated in this region by the THz waves absorption. Therefore, the variation of the resistance is maximum in the gap between a pair of tapered electrodes located at the center of this region. Figure 3(b) shows the back side of the fabricated metamaterial bolometer covered by the CB-PDMS layer combined with a pair of tapered gold electrodes. A pair of thin metal wires was attached on the electrodes using the silver paste and epoxy. Because the patterning resolution is limited when a shadow mask is used on the rough surface of the CB-PDMS layer, the minimum gap between the electrodes was determined to be 0.2 mm as shown in the right panel of Fig. 3(b).

Figure 3(c) shows the reference spectrum of the THz waves irradiated from the source. Before measurement via the THz-TDS system was performed, the THz waves were spatially filtered by a metal plate with a square window whose area is the same as that of the aperture in Fig. 1(a). Using the reference THz waves, which have maximum intensity near 1 THz as shown in Fig. 3(c), the transmitted and reflected power from the metamaterial bolometer were measured. Then, the transmittance and reflectance were calculated by dividing them with the reference spectrum as shown in Fig. 3(d). For the measurement of the reflected THz waves, the module utilized to focus the collimated THz waves in the transmittance measurement was replaced with a module using a silicon beam splitter for the normal incidence of the THz waves [9]. A TPX lens is added in front of the metamaterial bolometer to focus the THz waves on it as well as to collect all of the reflected THz waves from it. A mirror array was used to guide the reflected THz waves to the detector. To calculate the reflectance using the measurement, a reference spectrum was measured that was reflected by a gold-coated mirror. Notably, in the time domain, THz waves reflected from the metal plate having the aperture of 1 $\textrm {mm}^2$ measured without a device under test (DUT) was subtracted from the total reflected waves measured with DUT, to calculate the reflectance accurately.

When the measured reflectance and transmittance in Fig. 3(d) are compared with those of the calculated results in Fig. 2(d), a large discrepancy between them can be observed. The discrepancy can be mainly ascribed to the imaginary part of the permittivity of CB-PDMS, which was evaluated using the measured sheet resistance. As PDMS may have covered most of the CB microparticles that may have sunk during the curing of the mixture for 25 minutes under a $100^\circ$C temperature in a laboratory oven, the sheet resistance measured by the surface contact probes may have been overestimated. In addition, the dielectric loss was neglected in the modeling of the imaginary part of the permittivity, which implies that the effective conductivity of CB-PDMS was underestimated. To account for these errors, the surface resistance used in the simulation was tuned. As a result, it was confirmed that the differences between the simulated and the measured transmittance decrease when the measured sheet resistance is multiplied by a constant value of 0.3, as shown in Fig. 3(d). In spite of the adjustment of the imaginary part of the permittivity, it was found that the correction does not affect the resonant frequency.

Based on Fig. 3(d), it can be found that the reflectance is lower than the simulation value. The discrepancies between the simulated and measured results may occur owing to the roughness of the CB-PDMS layer and the corrugation on the thin ${\textrm {SiO}}_2$ layer suspended on the aperture. These could cause wave scattering that may decrease the power received by the detector. In addition, an error of alignment between the aperture and the center of the THz beam can induce the discordance, which may occur when the measurement setting for the transmittance is changed to that for the reflectance. The resonant frequency shifted rightward from 1 THz due to a fabrication error with regard to the thickness of the CB-PDMS layer. Nonetheless, it was verified that the reflectance was efficiently reduced near 1 THz as well as the transmittance was almost vanished owing to absorption by the proposed metamaterial bolometer. Because there exist some factors that deteriorate the accuracy of the measurement as mentioned above, the measured absorption ($A$) that can be calculated as $A=1-|T|-|R|$ was not compared with a simulated one.

To confirm whether the proposed metamaterial bolometer can detect the THz waves via the resistance measurement, the resistance of the bolometric layer was monitored using a multimeter connected to the electrodes on the bolometric layer. Before the THz waves were irradiated on the bolometer chip, the resistance of the bolometric layer isolated in the air was measured in a laboratory. The chip was heated by the body heat of the experimenter during the setup of the chip and cooled by natural convection to room temperature. Figure 4(a) shows that the resistance of the bolometric layer has exponentially decreased to a certain value, indicating the convective cooling of the chip. Each marker and error bar shows the average and standard deviation of the resistance, respectively, for the interval of one minute around the marker. The resistances were normalized using the final average value. The standard deviation shows that the thermal noise from the environment can vary the resistance. Therefore, to detect the variation caused by the incidence of the THz waves clearly, the average change should be sufficiently larger than the standard deviation of the resistance.

 

Fig. 4. Measured normalized resistances of the bolometric layer showing (a) convective cooling of the bolometer chip, and (b) increment and saturation of the measurement with irradiation of the THz waves.

Download Full Size | PDF

Figure 4(b) shows the normalized resistance of the bolometer when the THz waves were irradiated on the bolometer chip located inside the chamber of the THz-TDS system. The resistance was normalized using the initial average resistance under the irradiation condition of the THz waves. As shown in Fig. 4(b), the resistance of the bolometer chip was obviously increased by the irradiation of the THz waves. Within 5 minutes, the resistance increased steeply to a value that is higher than the initial value by 1.02$\%$. Based on the THz pulse with $E_{pp}$ of 20 V/cm as well as simulated absorption in Fig. 3(d), the total energy absorbed by the metamaterial bolometer was estimated to be 0.88 µJ. To calculate it, the energy density absorbed by the metamaterial bolometer was calculated as 9.33$\times 10^{-17}$ J/$\textrm {mm}^2$ by integrating the spectrum of the THz pulse via Parseval’s theorem after multiplying it by the simulated absorption in Fig. 3(d). Because $E_{pp}$ was measured using the collimated Gaussian beam [33], the energy density was multiplied by the spot area 0.39 $\textrm {mm}^2$ of the collimated beam. Then, the repetition rate 80 MHz was multiplied to the absorbed energy to calculate an energy per one second, i.e., an absorbed power. The absorbed power was calculated as 2.93 nW. Therefore, the total energy absorbed by the detector could be estimated by multiplying the power and the duration time 5 minutes.

Based on the calculated absorbed energy, the responsivity of the metamaterial bolometer could be evaluated. When the THz wave was irradiated on the device located in a chamber at 22.2$^\circ$C, the resistance of the bolometric layer changed by 1.02$\%$. Since TCR is 2.52$\%$/K at 22.2$^\circ$C as shown in Fig. 1(a), we could estimate the temperature variation caused by the THz wave is 0.405 K by dividing the resistance change by the TCR value. With the energy absorbed by the metamaterial bolometer of 0.88 µJ, the responsivity could be calculated to be 1.16$\%$/µJ (based on the resistance change) and 0.46 K/µJ (based on the temperature change). Although the response time of the detector is quite slow compared with those of the recently reported all-dielectric [19] and metal platinum based multispectral [20] metamaterial detectors, the proposed method has a novelty that the THz wave can be simply detected by measuring variation in the resistance of the bolometric layer (here the conductive CB-PDMS substrate). It may allow directly detecting molecules by spreading them on the surface of the metallic pattern. The electric fields that resonate between the metallic patches as shown in Fig. 2(c) could be utilize to detect molecules by interacting with them and affecting the dielectric loss of the lossy substrate. To improve the responsvity as well as response time, it could be a solution that reducing the thickness of the CB-PDMS layer by adopting much finer CB particles that have the diameters under 1 µm [35]. The reduction in the thickness of the bolometer could improve the responsivity and reduce the response time because it decreases heat capacity of the bolometric layer [18]. Besides, a high power THz source such as the THz quantum cascade laser [20] could improve the responsivity and the response time.

The red dashed line in Fig. 4(b) shows the exponentially saturating curve fitted to the measurements, which shows the convergence of the resistance from the initial value to the final value that is higher by 0.71$\%$. Even though the standard deviation of the resistance inside the chamber is larger than that measured in the laboratory environment owing to air conditioning inside the chamber, it is clearly confirmed that the normalized resistance increases to a value higher than the standard deviation owing to the absorption of the THz waves at room temperature 22.2$^\circ$C. Based on the experimental verification, it can be concluded that the detection of molecules via the bolometric method, i.e., by measuring the change in the resistance subsequent to THz irradiation, is feasible. The most realizable scenario to detect molecules utilizing the proposed metamaterial bolometer is that it is combined with the slot antennas [8–10] that can maximize the molecule’s absorption cross section. The frequency selectivity could be improved using Fano resonance which can be achieved by breaking the symmetry of the metallic pattern array [36]. This remains as a potential future study.

4. Conclusions

A compact and highly sensitive THz bolometer was proposed and its feasibility was experimentally verified. By utilizing the CB-PDMS layer with TCR ranging from 1.88$\%/K$ to 3.11$\%/K$ as a bolometric layer and adding a metamaterial absorber on the bolometric layer to increase the absorption of the THz waves, the irradiated THz waves could be detected via the bolometric method. The maximum change in resistance caused by the THz waves was 1.02$\%$, which was much higher than the standard deviation caused by the background thermal noise. There is still room for improvement in terms of the use of a better thermosensitive material to achieve a higher TCR, optimizing the design of the metamaterial absorber based on an accurate determination of the permittivity of the thermosensitive material, optimizing the device configuration such as the gap between the electrodes to improve the sensitivity, and optimizing the measurement settings for high power THz waves. In addition, the thermal-insulation techniques such as the packaging for pyroelectric detectors [15] or metastructures for thermal superinsulation [37] can be used to isolate the bolometer from background thermal noises. A further improved bolometer chip could sensitively detect the minute variations in the power of the THz radiations caused by the interaction of the radiations with chemical or biological molecules whose fingerprints are in the THz band.