Open Access
Add to BibSonomy
Abstract
Two metamaterial sensors were designed to test three pesticide residues. The influences of the metamaterial structure, the analyte composition and volume on the sensitivity have been studied. The metamaterial field-enhancement ability has an important influence on the sensitivity within the high-concentration range, while the coincidence between the metamaterial resonant frequency and the analyte fingerprint peak plays a dominant role within the low-concentration range. These findings allow us to better understand the process and find a way to improve the sensitivity.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
In the past few years, terahertz (THz) spectroscopy has become a promising technology, enabling label-free, non-contact and non-destructive testing of chemical and biological substances [1,2]. In particular, the recently developed portable THz spectroscopy tool can detect and identify the object materials on-site with a high signal-to-noise ratio [3], which further promoted the development of terahertz detection technology. However, at present, due to the low power of the terahertz source, the detection sensitivity available is limited, whether for the quantitative or qualitative analysis. The combination of the metamaterial and terahertz time-domain spectroscopy systems is an effective way to improve the detection sensitivity. Many researchers have applied this combination scheme in the field of medical [4,5], food safety [6,7] and other fields.
As we know, the quality factor Q, sensitivity S and FOM (figure of merit) are the main parameters to evaluate the sensor performances. For example, Wang [8] designed one kind of sub-wavelength metal resonance ring on a flexible film. The sensitivity S around the frequency 1.014 THz can reach 243 GHz/RIU, Q is 14.2, and FOM is 3.3. Shen [9] designed a complementary Dirac semi-metal metasource structure, with a sensitivity S of up to 302.5 GHz/RIU, a quality factor Q of 87.6, and FOM of 19. The ultra-narrow-band perfect absorber designed by Geng [10] has a sensitivity S of 190.4 GHz/RIU at 1.79 THz, the quality factor Q of 474, and FOM of 38.8. So we can find that the metamaterial biosensors do improve the detection sensitivity in the terahertz range.
However, most of the researches mainly focused on the analysis on the comparison between the sensor design and the experimental results. Researches on the factors and mechanisms affecting the sensitivity of terahertz detection are still insufficient. In this paper, two metamaterial sensors are designed, and three pesticide components are tested with these two sensors. The factors and mechanisms affecting the detection sensitivity during the practical application were analyzed from three perspectives: the metamaterial structure, the difference between the characteristic peak of the metamaterial and the fingerprint peak of the analyte, and the volume of the analyte.
2. Research methods
Two metamaterial sensors (structure 1 and structure 2) were designed, and the base was 10 µm thick polyimide (PI) as shown in Fig. 1. The experimental flowchart is shown in Fig. 1(a). The incident THz wave penetrates from PI to the metal layer as shown in Fig. 1(b). The designed structures, the actual prepared structures under the optical microscope and their dimensions are shown in Figs. 1(c) and 1(d). From Fig. 1(c) and Fig. 1(d), we can observe the dimensional errors of the prepared structures, which will cause the resonance frequency deviation between the designed and the measured results, shown in the following context. During the research, we used finite difference time domain (FDTD) algorithm to perform full-wave numerical simulation, and conducted a numerical study on the spectral response of the designed metamaterial structure. The permittivity ɛ and tanδ (Dissipation factor) of PI layer were 3.1 and 0.05, respectively [4]. Open boundary conditions were set in the z-direction, and unit cell boundary conditions were set in the x and y directions. According to the design results, the metamaterial samples were prepared. The preparation processes were as follows. First, a PI layer was coated on a high-resistance silicon substrate, and baked to imidize the PI. Then, the metamaterial pattern was obtained by standard photolithography (ABM mask aligner), and then 20 nm chromium and 200 nm Au layers were deposited by FHR magnetron sputtering. Acetone and isopropyl alcohol were used for 5 min to peel off the specimen. Then the specimen was divided into 8 mm ${\times} $ 8 mm with a Disco cutter, soaked in HF for 2 min, rinsed three times with the deionized water, and dried through blowing nitrogen. The PI film was removed with tweezers. The steps of the photolithography process are shown in Fig. 2. The detailed processing steps and parameters are given in Appendix Appendix 1. The measurement system we use is THz-TAS7400_TS, which uses two femtosecond laser-pumped optical antennas with slightly different frequencies to generate and detect terahertz waves through asynchronous sampling. The effective frequency range that TAS7400 can detect is 0.2–2.5 THz. The frequency resolution can be 7.9 GHz and 1.9 GHz. During our researches, we choose the frequency resolution 1.9 GHz.
Fig. 1. (a) Schematic diagram of testing operation (b) Propagation of the THZ wave during detection. (c)-(d) The designed structures, the physical images under the optical microscope and their dimensions
In order to facilitate evaluating the influence of the analyte composition and the volume of the detected solution, we needed to prepare the solid samples and liquid samples. In the preparation of solid samples, in order to prevent excessive absorption, polyethylene powder was added as a dispersant. Chlorothalonil, pyraclostrobin, procymidone, and polyethylene powder were crushed and grounded (100 mesh sieve) into small particles. Then, these small particles were dried in a vacuum oven at a temperature of 50°C for one hour to remove moisture. We mixed these small particles (chlorothalonil, pyraclostrobin, and procymidone) with polyethylene powder in a ratio of 1:1 to prepare round flakes. (total mass about 100 mg). In order to mix the powder evenly, we carefully ground the mixture with an agate pestle and an agate mortar. The mixture was then pressed into a solid sample about 1 mm thick and 13 mm in diameter. Before the spectrum was collected, the solid sample was relaxed for 30 minutes, so as to stabilize the samples size. For the liquid sample preparation, 0 mg/L, 0.1 mg/L, 0.5 mg/L, 1 mg/L, 5 mg/L, 10 mg/L (chlorothalonil, pyraclostrobin, and procymidone) solution were prepared by dissolving an appropriate amount of the peticide (chlorothalonil, pyraclostrobin, and procymidone) powder in acetone. During the test process, each liquid sample is tested for every metamaterial array, and the measurement is performed 12 times. The average value of the above 12 spectra is taken as the spectrum result of the corresponding sample. Before the formal researches, we have taken one structure as an example, testing the resonant frequencies 7 times a day to verify the sensor’s testing repeatability and test the resonant frequencies for 5 consecutive days to verify its testing stability. The detailed information is given in Appendix Appendix 2. The results demonstrate the good testing repeatability and stability of the prepared sensors.
3. Analysis and discussion
The transmission spectra of structures 1 and 2 were simulated and experimentally measured, shown in Fig. 3. Due to the dimensional errors induced during fabricating process, there appear the deviations between the simulated and the experimental results. For structure 2, the deviation is relatively more obvious. We analyze that is due to the bigger gap size error 1.809 µm (5µm–3.191 µm), which influence the equivalent capacitance seriously. For structure1, there was a clear resonance peak at 1.47 THz, and for structure 1, there was a clear resonance peak at 1.08 THz, as shown in Figs. 3(a) and 3(b). It can be seen from Fig. 3(c) that the dielectric constant of the PI film measured in the experiment is around 3, which is close to the value 3.1 set in our simulation. The phenomenon that the dielectric constant in Fig. 3(d) turns from negative to positive is explained detailed in [11,12].
Fig. 3. (a) Transmission spectra obtained through simulation and experiment for structure 1. (b) Transmission spectra obtained through simulation and experiment for structure 2 (c)Dielectric properties of PI film (d) Dielectric properties when 0.2 µm Au is plated on PI film
In the simulation, under the condition of 1 µm thick analyte layer [7], we changed the refractive index of the analyte layer uniformly from 1–1.9, and the results obtained are shown in Figs. 4(a)–4(d). From Figs. 4(b) and 4(d), it can be seen that the theoretical sensitivities of structure 1 and structure 2 are 156 GHz/RIU and 106 GHz/RIU, respectively.
Fig. 4. (a) Transmittance spectra of structure 1 (b) The theoretical sensitivity of structure 1 (c) Transmittance spectra of structure 2(d) The theoretical sensitivity of structure 2
3.1 Influence of the metamaterial structure on the detection sensitivity
We used structures 1 and 2 to test different concentrations of chlorothalonil solutions, and the test results were shown in Fig. 5.
Fig. 5. (a) Transmission spectra of different concentrations of chlorothalonil solution for structure 1 (b) Transmission spectra of different concentrations of chlorothalonil solution for structure 2 (c) Frequency shifts of the transmission peaks of structure 1, 2 for different concentrations of chlorothalonil solution (The inset is a partial enlarged view of area I).
Figure 5(a) and Fig. 5(b) show the transmission spectra of chlorothalonil at different concentrations of 0 mg/L, 0.1 mg/L, 0.5 mg/L, 1 mg/L, 5 mg/L, and 10 mg/L for structures 1, 2 respectively. From Fig. 5(a) and Fig. 5(b), we can extract the frequency shifts of the resonance peaks for the different solutions. The obtained frequency shifts are given in Fig. 5(c). In Fig. 5(c), we divide the range into two regions: I (shaded area) and II. It is found that the frequency shifts of structure 2 within region I are generally greater than those of structure 1, while within region II, the frequency shifts of structure 1 are greater than those of structure 2. Accordingly, we obviously observed that, within region I, the minimum detection limit of structure 2 is smaller. This means that structure 2 can identify smaller concentrations of solutes. We also can observe another phenomenon: in the region II, the frequency shifts for the two structures decrease comparing those in region I. We analyze this may be caused the aggregation of the residual solute after the volatizing of the high concentration of chlorothalonil solution, or some other reasons needing to be verified in the future.
As we know that, frequency shifts are caused by the refractive index changing induced by the concentration variation of the solute. So through analyzing the frequency shift, we can obtain the solute concentration. Within region II, at the same concentrations, the frequency shifts obtained with structure 1 are generally greater than those of structure 2. This means that in region 2, with structure 1, we can obtain greater frequency shift per unit concentration. Or we can say that structure 1 facilitates the quantitative detection of analytes. As we have pointed out previously, the theoretical sensitivity of structure 1 is 156 GHz/RIU, and the theoretical sensitivity of structure 2 is 106 GHz/RIU. So we can find that, within region 2, the sensitivity of the frequency shift caused by concentration changes is consistent with the theoretical sensitivity. On the other hand, in region 1, the sensitivity of the frequency shift caused by the concentration change is opposite to the theoretical sensitivity. This reminds us that within the low concentration range, the factor mainly affecting the sensitivity is not the theoretical sensitivity of the structure, but other factors. We will do a detailed analysis later.
In order to further explore the influencing mechanisms of the sensitivity, we obtained surface current distribution and Z-axis magnetic field of structures 1 and 2 at their resonance frequencies 1.47 THz and 1.08 THz through numerical simulation, as shown in Fig. 6. Because the metamaterial structure has strong electrical resonance or magnetic resonance (depending on the structure), it has a local electromagnetic field enhancement effect. Resonance occurs at the resonance frequency, and thus a peak appears at the corresponding frequency point. We can use the surface current distribution diagram to explain the formation of the magnetic field diagram. For structure1, as shown in Fig. 6(a), there are four circulating currents (marked by black arrows) at the four corners to form an LC resonance. At the center, the two circulating currents (marked by blue arrows) overlap. Because the generated magnetic fields are in opposite directions, the currents cancel out. Then it forms the magnetic field distribution in Fig. 6(b). For structure 2, in Fig. 6(c), five circulating currents (indicated with the arrows with different color and number) form LC resonance. At its center position (marked by the black arrow and No. 1), because the current density is the largest, the generated magnetic field is the strongest, thus forming the magnetic field distribution in Fig. 6(d). In addition, under the same intensity of external field excitation, we can observe that the current density in Fig. 6(a) is significantly greater than that in Fig. 6(c) (The density of the red arrows in the figure), and the intensity of the magnetic field generated in Fig. 6(b) is also stronger than that in Fig. 6(d). That is to say, the local enhancement effect of structure 1 is stronger than structure 2.
Fig. 6. (a)-(b) are the surface current distribution and Z-direction magnetic field of structure 1 at 1.47 THz; (c)-(d) are the surface current distribution and Z-direction magnetic field of structure 2 at 1.08 THz.
According to the formula of the resonance frequency:
where L and C are the equivalent inductance and capacitance of the structure.The expression of equivalent inductance L and equivalent capacitance C are given as [13]
where l and w are the length and width of path the current flowing through, h is the thickness of the intermediate substrate medium (in our experiment h is a fixed value of 10µm), and ${\varepsilon _r}$ is the relative permittivity of the medium. So [14],3.2 Influence of the analyte composition on the detection sensitivity
We mixed these small particles (chlorothalonil, pyraclostrobin, and procymidone) with polyethylene powder in a ratio of 1:1 to prepare round flakes. (total mass about 100 mg). With the THz-TDS test system, the time-domain spectra of the three pesticides were obtained, and the dielectric constant and transmission spectra of the three pesticides at 0.2–2 THz were obtained [15], as shown in Fig. 7. Figures 7(a)–7(b) shows the transmission spectra of metamaterial structure 2 respectively detecting different concentrations of pyraclostrobin and procymidone solutions. Figure 7(c) shows the frequency shifts obtained by detecting different concentrations of pesticides with structure 2.
Fig. 7. (a) Transmission spectra when detecting different concentrations of pyraclostrobin solution with structure 2 (b) Transmission spectra when detecting different concentrations of procymidone solution with structure 2 (c) Frequency shifts of the transmission peaks for different concentrations of pesticides (d) Permittivity of three pesticide components under THz time domain spectrum.
From Fig. 7(d), chlorothalonil has the largest dielectric constant, and procymidone has the smallest one. In Fig. 7(c), the detection sensitivity of chlorothalonil is the lowest, and the detection sensitivity of procymidone is the highest. So we conclude that the larger the dielectric constant, the smaller the frequency shift and the lower the sensitivity.
On the other hand, from Fig. 7(c), we can observe another phenomenon. Within the dashed box, for three pesticides, the difference of the detection limits and the line slope (indicating the ratio of the frequency shift and the concentration) of are not so obvious. About this phenomenon, we analyze that it is because within the low concentration range, the coincidence degree between the substance fingerprint peak and metamaterial resonant frequency plays the dominant role. As we all know that each substance has its unique fingerprint spectrum, and the metamaterial structure we designed also has its unique characteristic peaks. If the fingerprint resonance peak of the detected substance is close to the characteristic peaks of the metamaterial structure, the external field can interact with the substance more effectively through metamaterial-induced field enhancement. We detected the characteristic fingerprint peaks of the three pesticide components as shown in Figs. 8(a)–8(c), and plotted the comparison of the characteristic fingerprint peaks of the three pesticide components with the characteristic peaks of structure 1, 2 as shown in Table 1. It is found that within 0–2THz, the chlorothalonil has 3 distinct characteristic peaks, pyraclostrobin has 7 distinct characteristic peaks, and procymidone has 4 distinct characteristic peaks. The characteristic peaks of structure 1 and structure 2 are at 1.47 THz and 1.08 THz, respectively. It can be seen from Table 1 that around the characteristic peak of structure 2 at 1.08 THz, three pesticides all have fingerprint peaks, at 1.05THz, 1.06THz, and 1.08THz respectively. Therefore, it can be seen from Fig. 7(c) that within the low-concentration detection range, the minimum detection amounts of the three pesticides are almost the same, with a frequency shift of 8–13 GHz at about 0.1 mg/L. At the same time, the slope (indicating S value) are also nearly same. This result verifies that, within the low concentration range, the coincidence degree between the substance fingerprint peak and metamaterial structure characteristic peak plays the dominant role.
Fig. 8. (a) The absorption spectrum of Chlorothalonil solid (b) The absorption spectrum of Pyraclostrobin solid (c) The absorption spectrum of Procymidone solid
Table 1. Characteristic peaks of three pesticide components and two metamaterial structures
From Fig. 5(c), we also can find this mechanism. Although structure 1 has a stronger field-enhancement ability at 1.47THz than that of structure 2 at 1.08THz, chlorothalonil has the resonance peak at 1.05THz much closer to 1.08THz of structure 2. So in Fig. 5(c), within region I, structure 2 has the stronger ability to distinguish the substance and can detect the lower substance limit. That is to say, within the low concentration range, the smallest amount that can be detected by structure 2 is less than structure 1. This phenomenon further verifies that, within the low concentration range, the coincidence degree between the substance fingerprint peak and metamaterial structure characteristic peak plays the dominant role.
In summary, combined with the analysis we put forward in “3.1 The influence of metamaterial structure on the detection sensitivity,” within the high concentration detection range, the field enhancement ability of the structure has a decisive influence on the detection sensitivity. Within the low concentration detection range, the coincidence degree between the characteristic peak of the structure and the fingerprint peak of the analyte plays a dominant role in the detection of the minimum amount and the sensitivity S value. Or in other words, the high peak-to-peak coincidence is beneficial for qualitatively determining the presence or absence of the analyte, and has an important impact on the quantitative detection at low concentrations. At high concentrations, the field enhancement ability of the metamaterial sensor is beneficial for improving the sensitivity of quantitative detection.
3.3 Influence of the solution volume on the detection sensitivity
We tested the 10 mg/L chlorothalonil solution of different amounts for structure 2, and the obtained transmission spectrum and frequency shift diagram are shown in Fig. 9. During the experimental operation, we successively added 10µL, 10µL, 20µL, 20µL, 20µL, 10 mg/L chlorothalonil solution. We can find that the more the amount of dripping, the more the solute added (acetone is very volatile, and the dropped solution is equivalent to leaving only the solute on the surface of the metamaterial), the greater the value of the frequency shift.
Fig. 9. (a) Transmission spectra detected at different dripping amounts (b) Frequency shifts detected at different dripping amounts
The thickness of the analyte taken in this manuscript is 1 µm. As shown in Table 2, the sensitivities of structure 1 and structure 2 are 156 GHz/RIU and 106 GHz/RIU, respectively. Sensitivity is closely related to the thickness of the analyte. So if the thickness of the analyte for structure 1 is increased to 9 µm, according to Ref. [7], the sensitivity can reach to 367 GHz/RIU, which easily exceeds the sensitivity given in Table 2. In Refs. [8–10], the analyte thicknesses were not given. The Q and FOM of the two structures studied in this manuscript are lower than those in the literatures [8–10]. High Q value and FOM make the resonance peak sharper, which makes it easier to observe the resonance peak shift. In future, we can continue to optimize the metamaterial structure, so as to improve Q, FOM and sensitivity simultaneously.
Table 2. Comparison of sensor performance
4. Conclusion
In this paper, two types of THz sensors are designed. We used these two structures to detect three pesticide components, and analyzed the factors and the mechanism that affect the detection sensitivity in practical applications from the three perspectives: the metamaterial structure, the degree of agreement between the characteristic peak of the metamaterial structure and the fingerprint peak of the test substance, and the solution volume.
The main conclusions are as follows:
- (1) Greater field strength, is beneficial for improving the sensor sensibility. So in the practical application, the metamaterial structure should be carefully designed to achieve the stronger local field distribution.
- (2) For the analyte with the higher dielectric constant, we can obtain the lower detection sensitivity. And increasing the solution volume, we can observe the greater frequency shift, and obtain the higher sensitivity.
- (3) When detecting the low concentration range, the coincidence between the metamaterial characteristic peak and the substance fingerprint peak is the main factor affecting the sensitivity. Within the high concentration ranges, the field enhancement ability of the structure is the main factor affecting sensitivity.
- (4) Most researchers are now studying metamaterial structures with high sensitivity, high Q value, and low FOM value, but few people have explored the mechanisms underlined the sensitivity. The influences of the metamaterial structure, the analyte composition and volume on the sensitivity have been studied. The findings allow us to better understand and find the way to improve the sensitivity.
Appendix 1. Lithography steps
| Name | Condition | Time |
|---|---|---|
| Substrate | Four-inch high-resistance silicon | |
| Spin-on Coating | Positive rotation 600r/min | 5s |
| Counter rotation 800r/min | 30s | |
| Pre-bake | 120°C | 20 min |
| Imidize the PI | 120°C | 1h |
| 200°C | 2h | |
| 250°C | 3h |
| Name | Condition | Time |
|---|---|---|
| Spin-on Coating | Positive rotation 600r/min | 5s |
| Counter rotation 4000r/min | 30s | |
| Pre-bake | 95°C | 90s |
| Pre-exposure | 9.9mw/cm2 | 4.5s |
| Hard bake | 110°C | 60s |
| Flood exposure | 9.9 mw/cm2 | 45s |
| Development | TMAH3038 | 45s |
Acetone and isopropyl alcohol were used for 5 min to peel off the specimen.
Soked in HF for 3 min.| Name | Instrument | Size |
|---|---|---|
| Scribing | Disco cutter (hard knife) | 8mm×8mm |
Appendix 2. Repeatability and stability of our sensor
We have done some experiments to verify the repeatability and stability of our sensor.
Structure 2 was taken as an example. Without any solution dripping, the structure 2 is tested 7 times continuously, as shown in Table 3. The mean value of the resonance frequency value is 1.04196, and its standard deviation is 0.00377. It can be seen that the testing repeatability for the sensor is very stable.
Table 3. Seven consecutive tests
Then we tested the low-concentration 0.1 mg/L chlorothalonil solution seven times for five consecutive days, and measured the resonance frequency value of the transmission spectrum seven times a day, and the obtained resonance frequency-day changes are shown in Table 4. It can be seen that the standard deviation of the resonant frequency value of continuous testing for five days is 0.0018, indicating that our sensor has good stability.
Table 4. Test for five consecutive days
Funding
Six Talent Peaks Project in Jiangsu Province (GDZB-020).
Disclosures
No conflict of interest exits in the submission of this manuscript, and the manuscript is approved by all authors for publication. I would like to declare on behalf of my co-authors that the work described was the original research that has not been published previously, and not under consideration for publication elsewhere, in whole or in part.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. B. S. Alexandrov, V. Gelev, A. R. Bishop, A. Usheva, and K. O. Rasmussen, “DNA Breathing Dynamics in the Presence of a Terahertz Field,” Phys Lett A 374(10), 1214–1217 (2010). [CrossRef]
2. H. R. Park, K. J. Ahn, S. Han, Y. M. Bahk, N. Park, and D. S. Kim, “Colossal absorption of molecules inside single terahertz nanoantennas,” Nano Lett 13(4), 1782–1786 (2013). [CrossRef]
3. P. Zhao, S. Ragam, Y. J. Ding, and I. B. Zotova, “Compact and portable terahertz source by mixing two frequencies generated simultaneously by a single solid-state laser,” Opt. Lett. 35(23), 3979 (2010). [CrossRef]
4. X. Yan, M. Yang, Z. Zhang, L. Liang, D. Wei, M. Wang, M. Zhang, T. Wang, L. Liu, J. Xie, and J. Yao, “The terahertz electromagnetically induced transparency-like metamaterials for sensitive biosensors in the detection of cancer cells,” Biosens Bioelectron 126, 485–492 (2019). [CrossRef]
5. Z. Zhang, M. Yang, X. Yan, X. Guo, J. Li, Y. Yang, D. Wei, L. Liu, J. Xie, Y. Liu, L. Liang, and J. Yao, “The Antibody-Free Recognition of Cancer Cells Using Plasmonic Biosensor Platforms with the Anisotropic Resonant Metasurfaces,” ACS Appl Mater Interfaces 12(10), 11388–11396 (2020). [CrossRef]
6. P. Nie, D. Zhu, Z. Cui, F. Qu, L. Lin, and Y. Wang, “Sensitive detection of chlorpyrifos pesticide using an all-dielectric broadband terahertz metamaterial absorber,” Sensors and Actuators B: Chemical 307, 127642 (2020). [CrossRef]
7. Y. Zhang, Y. Ye, X. Song, M. Yang, Y. Ren, X. Ren, L. Liang, and J. Yao, “High-sensitivity detection of chlorothalonil via terahertz metasensor,” Mater. Res. Express 7(9), 095801 (2020). [CrossRef]
8. Z. Wang, Z. Geng, and W. Fang, “Exploring performance of THz metamaterial biosensor based on flexible thin-film,” Opt Express 28(18), 26370–26384 (2020). [CrossRef]
9. S. Shen, Y. Liu, W. Liu, Q. Tan, J. Xiong, and W. Zhang, “Tunable electromagnetically induced reflection with a high Q factor in complementary Dirac semimetal metamaterials,” Materials Research Express 5 (2018).
10. Z. Geng, W. Su, X. Wang, Y. Jiang, and Y. Liu, “Numerical design of a metasurface-based ultra-narrow band terahertz perfect absorber with high Q-factors,” Optik 194, 163071 (2019). [CrossRef]
11. P. Xie, Z. Zhang, Z. Wang, K. Sun, and R. Fan, “Targeted Double Negative Properties in Silver/Silica Random Metamaterials by Precise Control of Microstructures,” Research 2019, 1–11 (2019). [CrossRef]
12. H. Luo and J. Qiu, “Carbon nanotubes / epoxy resin metacomposites with adjustable radio-frequency negative permittivity and low dielectric loss,” Ceram. Int. 45(1), 843–848 (2019). [CrossRef]
13. X. Guoqing, “Study on the metamaterial absorber and its sensing application,” (Southwest University, 2018).
14. X. Yan, X. Zhang, L. Liang, and J. Yao, “Research progress on the application of terahertz-band metamaterials in biosensors,” Spectroscopy and Spectral Analysis 34, 2365–2371 (2014).
15. J. Neu and C. A. Schmuttenmaer, “Tutorial: An introduction to terahertz time domain spectroscopy (THz-TDS),” Journal of Applied Physics 124 (2018).
