Identification and quantitative detection of two pathogenic bacteria based on a terahertz metasensor†

Zhaofu Ma‡ ^ab, Yanan Jiao‡ ^ab, Chiben Zhang‡ ^c, Jing Lou ^bc, Pengyue Zhao ^a, Bin Zhang ^a, Yujia Wang ^a, Ying Yu ^c, Wen Sun ^d, Yang Yan ^a, Xingpeng Yang ^a, Lang Sun ^b, Ride Wang ^b, Chao Chang *^b, Xiru Li *^a and Xiaohui Du *^a
aDepartment of General Surgery, First Medical Center, Chinese PLA General Hospital, Beijing 100853, China. E-mail: duxiaohui301@sina.com; 2468li@sina.com
^bInnovation Laboratory of Terahertz Biophysics, National Innovation Institute of Defense Technology, Beijing 100071, China. E-mail: gwyzlzssb@pku.edu.cn
^cAir and Missile Defense College, Air Force Engineering University, Xi'an 710051, China
^dDepartment of Anesthesiology, The Second Affiliated Hospital, Tianjin University of Traditional Chinese Medicine, Tianjin 300250, China

First published on 29th November 2022

---

---

1 Introduction

As a novel optical detection technology, terahertz (THz) waves have been widely applied in the biomedical field owing to their excellent characteristics, such as low photon energy and strong penetrability.^1–3 Considering the fingerprint frequency of biomolecules occurring in the THz band, THz spectroscopy provides a new platform for the qualitative and quantitative detection of biological macromolecules including DNA, RNA, amino acids, and proteins.^4–6 However, the lack of powerful radiation sources and the quite weak inter- or intra-molecular vibrations of the analyte in the THz region significantly limit the development of the THz technology for biological detection applications. To fully amplify biological information for wider application of THz technology in the biomedical field, a thought-provoking way is to strengthen the coupling effect between analytes and THz waves. Fortunately, such encountered bottlenecks can be addressed by combining resonant metasurfaces with THz technology, with the merits of high sensitivity and portable operation.⁷

Metasurfaces are planarized, ultra-thin, and artificial two-dimensional structures composed of subwavelength meta-atoms or nano-antennas arranged periodically, offering novel effects by controlling the response that originates from the interaction between meta-atoms and electromagnetic radiation.^8–10 The remarkable functionalities of metasurfaces such as the negative refraction index and invisibility cloak have attracted a great deal of attention from researchers in various fields.^2,11–16 With the rapid progress in recent decades, metasurfaces have achieved ground-breaking developments in applications such as achromatic meta-lenses, lasing spasers, and polychromatic holography, among others, and have rapidly permeated various branches of modern nanophotonics.^17–25 Metasurfaces offer an excellent platform to achieve enhanced light–matter interactions by controlling the optical responses in the desired manner.^26–30 Due to the robust near-field enhancement, extremely sensitive metasurfaces are employed as optical sensors to perceive minor surrounding changes.^31,32 For instance, ultra-sensitive THz metasensors have been proposed to distinguish extremely diluted concentrations of glucose, interleukin-6 (IL-6) and other biomolecular solutions.^33–36 Furthermore, the continuous optimization of the THz metasensors enables complex cell detection, including monitoring the carcinogenesis process of normal cells and discriminating the molecular typing of tumour cells, which is a major advance in the field of biosensing.^37–39 Arbitrary radiating metasurfaces can be analysed by an array of oscillating charge and loop currents, which involves the traditional concept of electric and magnetic multipoles, respectively. In recent years, the advent of a brand-new third family of electromagnetic multipoles, the toroidal dipole, has become equally important for the complete multipole analysis of metasurface radiation.^40–42 As the fundamental mode of toroidal moment, the toroidal dipole is more complex than the electric dipole and magnetic dipole. The toroidal response on the metasurface can be aroused by logically configuring the symmetry of split-ring resonators (SRRs) and their spatial arrangement. The unique light field localization of the toroidal response arises from its weak or non-radiative characteristics, which enables the resonator to store more energy and reduce the energy loss rate.⁴³ Among various designs of metasurfaces, artificial metal–dielectric metasurfaces based on constructive and destructive interferences generated by electric and toroidal dipole resonances have conspicuous physical properties. Two kinds of dipoles can usually be constructed, namely bright mode and dark mode, with different Q factors, and the coupling of the bright–dark modes generates electromagnetically induced transparency (EIT).^44–46 Compared to the traditional resonances induced by a single dipole, the coherence resonances based on the cooperative effect of two dipoles usually have a sharper line shape and higher Q factor, which render them suitable for sensing applications.^32,47,48

Bacterial culture, the most common clinical method, is the “gold standard” for diagnosing bacterial infection. However, this method requires one or two days to culture the bacteria to the concentration required for testing, and the bacteria are identified by a professional microbiologist. Other bacterial detection methods, such as labelled immunoassays, flow cytometry, and fluorescence microscopy, remain important in clinical microbiology laboratories when sample volumes are large and the patient's condition is not critical.^49–53 Nevertheless, highly sensitive, low-cost and label-free optical sensors based on bioanalyte recognition and measurement of refractive index changes are on the research agenda for the diagnosis of infectious diseases, which will be effective in identifying pathogens and the extent of infection, thus assisting doctors to formulate more accurate therapeutic schedules.

In this work, based on the coupling of the electric dipole and the toroidal dipole, a coherent resonance is comprehensively demonstrated to design a THz metasensor. To better explain the interference between the electric dipole and the toroidal dipole, we discuss the physical mechanism behind the excitation of resonance in detail and give the corresponding simulations. Due to the enhanced effect of the toroidal dipole on the electric dipole, the refractive index sensitivity and Q factor of the metasurface reach as high as 378 GHz per refractive index unit (RIU) and 21.28, respectively. In addition, a unique structure is discovered to render high biosensing capability by boosting light–matter interactions, which contributes to bacterial identification and quantitative detection. Thus, this work demonstrates the practical application of THz metasurfaces and will further advance the development of THz sensors in the biomedical field.

2 Materials and methods

Fig. 1(a) shows the schematic diagram of the THz metasurface biosensing. The THz beam is perpendicular to the metasensor with its polarization parallel to the cut wire (CW). The nuances of the analyte covering on the metasensor are reflected in the THz transmission spectra, enabling the identification and quantitative detection of bacteria. Fig. 1(b) reveals the detailed structural design of the unit meta-atom that consists of an isolated metal CW and a pair of split-ring resonators. An individual cell is a square with a period P = 75 μm in both the x and y directions. The length of the CW arranged in the middle of one lattice is L = 50 μm and a pair of SRRs are symmetrically located on both sides of the CW with a space of b = 2 μm. The length of the SSRs is a = 40 μm, the radius of the semicircle is r = 6 μm, and the split-gap size is c = 2 μm. The arm width of the metal structures is w = 4 μm. The preparation of metasensors adopted the standard micro/nanomachining technology. First, a 500 μm-thick quartz substrate was cleaned with acetone and isopropyl alcohol to remove the surface dust. Next, the metal structures were prepared by step lithography and ion beam etching methods on 10 nm-thick titanium and 160 nm-thick gold films. In the metal structure of the metasurface is a 10 nm-thick titanium adhesive layer that tightly holds the 160 nm-thick gold and the 500 μm-quartz substrates together. The effect of the 10 nm-thick titanium layer on resonance is shown in Fig. S1, ESI.† According to the microscopy results, the etching precision along the lateral and longitudinal directions is ±0.5 μm and +10%, respectively. The ultimate size of the prepared metasensor is 1 × 1 cm², with 17689 cells arranged periodically along the x and y directions and in a single layer along the z direction. The optical image of the prepared metasensor can be seen in Fig. 1(c).

Two parts of the unit cell include a metal CW and a pair of SRRs, a control electric dipole and a toroidal dipole, respectively. For an electric dipole resonator, as shown at the top of Fig. 2(a), current (c₁, blue line) flows from a positive charge to a negative charge, thereby creating a magnetic field (m₁, red line) around the metal CW. As shown at the bottom of Fig. 2(a), the surface current (c₂, blue line) on the symmetrical SSR flows in the opposite direction, inducing a pair of reverse magnetic dipoles. The magnetic dipoles are further coupled to form a circular magnetic field (m₂, blue line), which is confined in a small space and tightly around the CW. According to the right-hand rule, the direction of the toroidal dipole (T, yellow line) is parallel to the x-direction. The toroidal dipole can localize the energy by concentrating the time-varying magnetic field in a small circular region, thus greatly restraining the radiation loss between the structure and free space, and enhancing light–matter interactions. The toroidal dipole reinforces the electric dipole when the magnetic field generated by the toroidal dipole resonator is in the same direction as the magnetic field generated by the metal CW. In the opposite direction, the electric dipole is weakened. Therefore, mode I and mode II of the resonance are formed by the enhancement and weakening of the toroidal dipole, respectively.

To validate the response of the designed metasensor, numerical simulation based on the Computer Simulation Technology (CST) Microwave Studio 2018 was employed. Periodic boundary conditions were employed along the x-direction and y-direction and the perfectly matched layer (PML) absorbing boundary condition was applied in the z-direction in the free space. The incident wave was a polarized wave with its polarization parallel to the CW (x-direction) and perpendicular to the unit-cell along the z-direction. The electric field was along the x-direction. Each wavelength corresponded to 20 meshes with tetrahedral subdivision. In the simulation, the dielectric constant of the quartz substrate was 3.58. The conductivity of Au was 4.56 × 10⁷ S m⁻² and that of Ti was 1.8 × 10⁷ S m⁻². As shown in Fig. 2(b), we simulated the transmission spectra of the metasensor and clearly identified two resonance dips, which were labelled as mode I and mode II, highlighted in the red and blue areas, respectively. The Q factor of a resonator commonly indicates the rate of energy loss relative to the stored energy in the resonator. In addition, the resonance intensity is a key indicator of sensing performance in practical applications. The constructive and destructive hybridizations of two dipoles with different lengths of CW are simulated, as shown in Fig. 2(b). According to the following formulas:

ΔI = Ipeak − Idip

where f₀ represents the central frequency of the resonant peak and FWHM represents the full width at half maximum/minimum at the resonance. Fig. 2(c) shows the variation of the Q factor as the length of CW increases for the two modes. The Q factor for mode I increases and then decreases as the CW lengthens, reaching a maximum value when L = 40 μm. As for mode II, the Q factor increases without a downward trend. The improvements in the resonance intensity for two modes as the length of the CW increases are indicated in Fig. 2(d). The resonance intensity is pretty weak when the length of the CW is set to 40 μm, although the Q factor reaches its maximum. When stretching the CW to 70 μm, the Q factor of mode I is too weak to meet the actual sensing requirements. Based on the above simulation results, the length of the CW was determined to be 50 μm to obtain the trade-off between the Q factor and the resonance intensity. In addition, the effect of other structural parameters on resonance, such as the cut size of the SRRs and the distance between the SSRs and the CW can be seen in Fig. S2, ESI.† To further substantiate the effect of coupling, we simulated the surface current distribution, E-field intensity confined around the metallic structure and the z-component H-field confinement for both modes, as shown in Fig. 3. The strong current distribution on the CW pronounces a relatively stronger coupling for mode I (Fig. 3(a) and (c)). In addition, stronger surface currents indicate a high amount of energy confined in the metasensor array which is due to the enhancement of the toroidal dipole. Fig. 3(b), (c), (e) and (f) depict the stored electrical energy and magnetic energy in the metallic structure, which are also relatively stronger for mode I.

3 Results and discussion

To further explore the sensing properties of the metasensor, we regarded the analyte as a homogeneous dielectric with a thickness of 10 μm and numerically simulated the effects of its refractive index (n) and dielectric dissipation factor (tanδ) on the resonance as shown in Fig. 4(a) and (b). Fig. 4(c) shows the shift results in the case of identical tanδ but different refractive indices at the two modes. Without the covered analyte, mode I and mode II first occur at 1.522 THz and 2.252 THz, respectively. When varying the refractive index from n = 1 to n = 1.5 in steps of 0.1, the corresponding red shifts of mode I occur at 40 GHz, 86 GHz, 130 GHz, 160 GHz, and 190 GHz, respectively, and the red shifts of mode II are 30 GHz, 62 GHz, 92 GHz, 122 GHz, and 154 GHz, respectively. Based on the simulation results, we calculated the sensitivities of mode I and mode II as 378 GHz RIU⁻¹ and 305 GHz RIU⁻¹, respectively, according to the following formula:

where S (sensitivity) is expressed in GHz RIU⁻¹, Δƒ represents the frequency shift and Δn represents the amount of change in the refractive index. The normalized frequency of the THz metasensor was calculated to be 28.68%. When keeping n = 1.2 and varying the dielectric dissipation factor from tanδ = 0 to tanδ = 0.15 in steps of 0.03, the intensity of resonance gradually reduces and the frequency negligibly varies for the two modes as shown in Fig. 4(e). The significant factors including frequency shift and amplitude modulation are extracted to quantitatively analyse the sensing performance of the metasurface (Fig. 4(d) and (f)). For variations of the refractive index, subtle differences in the frequency shift for the two modes can be ignored. However, upon varying the dielectric dissipation factor, mode I exhibits a relatively significant modulation for amplitude compared to mode II. In summary, the enhancement of the electric dipole by the toroidal dipole enables mode I to possess a relatively stronger sensing capability in practice.

In order to explore the practical applications of the metasensor, a study of bacterial detection was carried out by taking Escherichia coli and Staphylococcus aureus as examples. In our experiment, two standard bacteria including Escherichia coli ATCC 25922 and Staphylococcus aureus ATCC 25923 were obtained from the National Center for Clinical Laboratories. Following the standard procedure of bacterial culture, the thawed strains were inoculated into Luria–Bertani solid medium and placed in a bacterial incubator (37.0 degrees celsius and 5% carbon dioxide, thermo) to form bacterial colonies. After 24 hours, individual colonies were picked and subcultured again in Luria–Bertani solid medium. The two subcultures enabled the strains to maintain stable living conditions. Next, the bacterial colonies were uniformly suspended in deionized water (to avoid the influence of salt crystallization on the results after water evaporation) and the McIntosh turbidity was adjusted to 0.5 MCF (approximately equal to 10⁸ cfu mL⁻¹). Finally, the bacterial suspension was diluted to the desired concentrations, including 5 × 10⁴ cfu mL⁻¹, 1 × 10⁵ cfu mL⁻¹, 5 × 10⁵ cfu mL⁻¹, and 1 × 10⁶ cfu mL⁻¹, so as to obtain 8 samples for experimental detection. 100 μL of the sample was pipetted to evenly cover the metasensor and then placed in a drying oven set at a temperature of 40 °C. Micrographs of the samples with different concentrations placed on the metasensor are clearly shown in Fig. S3, ESI.† After the samples were dried, the metasensor covered with bacteria was taken out of the oven and placed on the detection platform of the THz time-domain spectroscopy (THz-TDS) system. The THz-TDS system was wrapped in a closed acrylic box and equipped with a drying unit to control the humidity conditions required for detection. Finally, the metasensor covered with bacteria was probed using the THz-TDS system. The metasensor was cleaned thoroughly before adding the next sample and the 8 samples were analysed in sequence as described above.

In this study, all the detections of the samples were performed at room temperature (25 °C) and under a dry environment (1.5% humidity) to avoid the influence of water vapor. A THz beam with a 2 mm diameter spot illuminated the metasensor with a size of 1 × 1 cm². Fig. 5(a) shows the THz transmission spectra of Escherichia coli at different concentrations and the insets reveal enlarged views of the violet wireframes. We added the spectra of a bare metasensor (black lines in Fig. 5(a) and (b)) in order to compare the influence of bacteria on the resonance. When increasing the concentrations of Escherichia coli, the resonance shifts to a low frequency and the resonance intensity gradually diminishes. The red shifts of the resonance occur at 3.81 GHz, 11.44 GHz, 22.88 GHz, and 30.51 GHz, respectively, and the relative amplitude modulations are 0.0013, 0.024, 0.032, and 0.042, respectively. Similarly, when increasing the concentration of Staphylococcus aureus, the corresponding red shifts of resonance occur at 15.25 GHz, 19.07 GHz, 30.51 GHz and 41.96 GHz, respectively, and the relative amplitude modulations are 0.015, 0.034, 0.061 and 0.089, respectively, as shown in Fig. 5(b). The experimental results are consistent with previous simulation results. In addition, we plot the frequency shift (Δf) and relative amplitude modulations (Δa) against the bacterial concentrations, respectively, as shown in Fig. 5(c) and (d). A similar trend can be observed in the graph for both bacteria but the two bacteria have different rates of change in amplitude modulation and frequency shift with the same variations of concentration. Mode I has more evident frequency shifts and amplitude modulations than mode II when parameters including the refractive index and the dielectric dissipation factor of the bacterial analyte are varied in the practical experiment. The minimum detectable concentration for both the bacteria is as low as 5 × 10⁴ cfu mL⁻¹ and four concentrations of bacterial suspensions can be quantified. The analyses and discussions about Mode II are shown in Fig. S4, ESI.† The example proves the feasibility of using the designed metasurface as a biosensor for bacterial detection to develop real-time and portable THz devices for inspecting biological and chemical trace substances.

4 Conclusions

In this work, the designed artificial metal–dielectric metasensor based on constructive and destructive interferences that are generated by electric dipole and toroidal dipole coupling has conspicuous physical properties that include a high Q factor and robust resonance intensity. The metasensor has frequency-domain sensing ability due to the strong field confinement with a high sensitivity of 378 GHz RIU⁻¹, which is beneficial for detecting more subtle information and thus sensing bacterial solutions with concentrations down to ∼10⁴ cfu mL⁻¹. Experiments regularly demonstrate variations in frequency and relative amplitude with increases in bacteria concentration, which is consistent with our simulation results. In addition, the relative amplitude modulation and frequency shift information of the two bacteria at different concentrations were extracted and analysed from the experimental data, enabling the identification and quantitative detection of Escherichia coli and Staphylococcus aureus. The proposed high Q factor metasensor based on the coupling of electric and toroidal dipoles extends the study of resonant coherence effects in biosensing and opens up an effective pathway for early and rapid label-free biomedical detection.

Author contributions

Conceptualization, writing – original draft preparation, and visualization, Zhaofu Ma, and Yanan Jiao; software and formal analysis, Chiben Zhang; methodology and investigation, Jing Lou; validation, Pengyue Zhao; data curation, Bin Zhang; writing – review and editing, Yujia Wang, Ying Yu, and Wen Sun; supervision, Yang Yan and Xingpeng Yang; resources, Lang Sun, and Ride Wang; project administration, Xiaohui Du and Xiru Li; funding acquisition, Xiaohui Du and Chao Chang. All authors have read and agreed to the published version of the manuscript.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This research was supported by the National Natural Science Foundation of China (81871317, 12225511, and T2241002). C. C. acknowledges the XPLORER Prize no. 2020-1023.

References

1. G. Liu, C. Chang, Z. Qiao, K. Wu, Z. Zhu, G. Cui, W. Peng, Y. Tang, J. Li and C. Fan, Adv. Funct. Mater., 2019, 29, 1807862 CrossRef .
2. T. Yen, W. Padilla, N. Fang, D. Vier, D. Smith, J. Pendry, D. Basov and X. Zhang, Science, 2004, 303, 1494–1496 CrossRef CAS PubMed .
3. D. Suzuki, S. Oda and Y. Kawano, Nat. Photonics, 2016, 10, 809–813 CrossRef CAS .
4. B. Fischer, M. Hoffmann, H. Helm, R. Wilk, F. Rutz, T. Kleine-Ostmann, M. Koch and P. Jepsen, Opt. Express, 2005, 13, 5205–5215 CrossRef CAS PubMed .
5. J. Yu, C. Huang, Y. Sun, X. Su, H. Cao, J. Hu, K. Wang, J. Suo, K. Tao, X. He, H. Wei, M. Ying, W. Hu, X. Du, Y. Hu, H. Liu, C. Zheng, P. Li, J. Xie, F. Liu, Z. Li, G. Zhao, K. Yang, C. Liu, H. Li, P. Chen, J. Ji and G. Li, J. Am. Med. Assoc., 2019, 321, 1983–1992 CrossRef PubMed .
6. M. B. Díaz and J. J. López, Adv. Opt. Mater., 2020, 8, 1900721 CrossRef .
7. I. J. S. Nikolay, Science, 2010, 328, 582–583 CrossRef PubMed .
8. P. Pitchappa, A. Kumar, R. Singh, C. Lee and N. Wang, J. Micromech. Microeng., 2021, 31, 113001 CrossRef CAS .
9. Y.-S. Lin and Z. Xu, Int. J. Optomechatronics, 2020, 14, 78–93 CrossRef .
10. C. Zhang, G. Wang, H. X. Xu, X. Zhang and H. P. Li, Adv. Opt. Mater., 2020, 8, 1901719 CrossRef CAS .
11. A. Popov and V. Shalaev, Opt. Lett., 2006, 31, 2169–2171 CrossRef PubMed .
12. D. Schurig, J. J. Mock, B. Justice, S. A. Cummer, J. B. Pendry, A. F. Starr and D. R. Smith, Science, 2006, 314, 977–980 CrossRef CAS PubMed .
13. J. B. Pendry, D. Schurig and D. R. Smith, Science, 2006, 312, 1780–1782 CrossRef CAS PubMed .
14. W. Cai, U. K. Chettiar, A. V. Kildishev and V. M. Shalaev, Nat. Photonics, 2007, 1, 224–227 CrossRef CAS .
15. K. Shih, Z. Ren, C. Wang and C. Lee, J. Phys. D: Appl. Phys., 2019, 52, 394001 CrossRef CAS .
16. C. Zhang, X. Zhang, P. Xie, H. Hou, K. Liu, J. Lou, H. Bai, J. Zhang, Z. Li, W. Zhu and G. Wang, ACS Appl. Electron. Mater., 2021, 3, 4465–4471 CrossRef CAS .
17. Q. Yang, J. Gu, D. Wang, X. Zhang, Z. Tian, C. Ouyang, R. Singh, J. Han and W. Zhang, Opt. Express, 2014, 22, 25931–25939 CrossRef PubMed .
18. A. She, S. Zhang, S. Shian, D. R. Clarke and F. Capasso, Sci. Adv., 2018, 4, eaap9957 CrossRef PubMed .
19. A. Kodigala, T. Lepetit, Q. Gu, B. Bahari, Y. Fainman and B. Kante, Nature, 2017, 541, 196–199 CrossRef CAS PubMed .
20. Z.-L. Deng, X.-C. Qiao, Y.-B. Chen, T. Shi, X. Li and J.-W. Dong, J. Phys. D: Appl. Phys., 2021, 54, 345101 CrossRef CAS .
21. W. T. Chen, K. Y. Yang, C. M. Wang, Y. W. Huang, G. Sun, I. D. Chiang, C. Y. Liao, W. L. Hsu, H. T. Lin, S. Sun, L. Zhou, A. Q. Liu and D. P. Tsai, Nano Lett., 2014, 14, 225–230 CrossRef CAS PubMed .
22. Y. W. Huang, W. T. Chen, W. Y. Tsai, P. C. Wu, C. M. Wang, G. Sun and D. P. Tsai, Nano Lett., 2015, 15, 3122–3127 CrossRef CAS PubMed .
23. Z. Huang, D. L. Marks and D. R. J. O. Smith, Optica, 2019, 6, 119–124 CrossRef CAS .
24. J. Lou, J. Liang, Y. Yu, H. Ma, R. Yang, Y. Fan, G. Wang and T. Cai, Adv. Opt. Mater., 2020, 8, 2000449 CrossRef CAS .
25. H. Zhou and X. Mu, Int. J. Optomechatronics, 2021, 15, 97–119 CrossRef .
26. N. Yu and F. Capasso, Nat. Mater., 2014, 13, 139–150 CrossRef CAS PubMed .
27. A. Ahmadivand, M. Semmlinger, L. Dong, B. Gerislioglu, P. Nordlander and N. J. Halas, Nano Lett., 2018, 19, 605–611 CrossRef PubMed .
28. J. Lou, X. Xu, Y. Huang, Y. Yu, J. Wang, G. Fang, J. Liang, C. Fan and C. Chang, Small, 2021, 17, e2104275 CrossRef PubMed .
29. Z. Ren, Y. Chang, Y. Ma, K. Shih, B. Dong and C. Lee, Adv. Opt. Mater., 2019, 8, 1900653 CrossRef .
30. K. Shih, P. Pitchappa, M. Manjappa, C. P. Ho, R. Singh and C. Lee, J. Appl. Phys., 2017, 121, 023102 CrossRef .
31. Y. Xu, Y. S. Ang, L. Wu and L. K. Ang, Nanomaterials, 2019, 9, 165 CrossRef CAS PubMed .
32. M. Gupta and R. Singh, Adv. Opt. Mater., 2016, 4, 2119–2125 CrossRef CAS .
33. R. Wang, L. Xu, J. Wang, L. Sun, Y. Jiao, Y. Meng, S. Chen, C. Chang and C. Fan, Nanoscale, 2021, 13, 18467–18472 RSC .
34. L. Sun, L. Xu, J. Wang, Y. Jiao, Z. Ma, Z. Ma, C. Chang, X. Yang and R. Wang, Nanoscale, 2022, 14, 9681–9685 RSC .
35. C. Tang, J. Yang, Y. Wang, J. Cheng, X. Li, C. Chang, J. Hu, J. J. S. Lü and A. B. Chemical, Sens. Actuators, B, 2021, 329, 129113 CrossRef CAS .
36. K. Shih, P. Pitchappa, L. Jin, C.-H. Chen, R. Singh and C. Lee, Appl. Phys. Lett., 2018, 113 Search PubMed .
37. C. Zhang, T. Xue, J. Zhang, L. Liu, J. Xie, G. Wang, J. Yao, W. Zhu and X. J. N. Ye, Nanophotonics, 2022, 11, 101–109 CrossRef CAS .
38. J. Lou, Y. Jiao, R. Yang, Y. Huang, X. Xu, L. Zhang, Z. Ma, Y. Yu, W. Peng, Y. Yuan, Y. Zhong, S. Li, Y. Yan, F. Zhang, J. Liang, X. Du, C. Chang and C. W. Qiu, Proc. Natl. Acad. Sci. U. S. A., 2022, 119, e2209218119 CrossRef PubMed .
39. Y. Jiao, J. Lou, Z. Ma, L. Cong, X. Xu, B. Zhang, D. Li, Y. Yu, W. Sun, Y. Yan, S. Hu, B. Liu, Y. Huang, L. Sun, R. Wang, R. Singh, Y. Fan, C. Chang and X. Du, Mater. Horiz., 2022, 9, 2984–2992 RSC .
40. N. Papasimakis, V. A. Fedotov, V. Savinov, T. A. Raybould and N. I. Zheludev, Nat. Mater., 2016, 15, 263–271 CrossRef CAS PubMed .
41. V. Savinov, V. A. Fedotov and N. I. Zheludev, Phys. Rev. B: Condens. Matter Mater. Phys, 2014, 89, 205112 CrossRef .
42. M. Gupta, Y. K. Srivastava and R. Singh, Adv. Mater., 2018, 30, 1704845 CrossRef PubMed .
43. R. Yang, J. Xu, N. H. Shen, F. Zhang, Q. Fu, J. Li, H. Li and Y. Fan, InfoMat, 2021, 3, 577–597 CrossRef CAS .
44. W. Yu, H. Meng, Z. Chen, X. Li, X. Zhang, F. Wang, Z. Wei, C. Tan, X. Huang and S. Li, Opt. Commun., 2018, 414, 29–33 CrossRef CAS .
45. S. Hu, D. Liu and H. J. Yang, J. Phys. D: Appl. Phys., 2019, 52, 175305 CrossRef CAS .
46. Y. Yang, I. I. Kravchenko, D. P. Briggs and J. Valentine, Nat. Commun., 2014, 5, 5753 CrossRef CAS PubMed .
47. P. Tassin, L. Zhang, T. Koschny, E. N. Economou and C. M. Soukoulis, Phys. Rev. Lett., 2009, 102, 053901 CrossRef CAS PubMed .
48. X. Yan, M. Yang, Z. Zhang, L. Liang, D. Wei, M. Wang, M. Zhang, T. Wang, L. Liu, J. Xie and J. Yao, Biosens. Bioelectron., 2019, 126, 485–492 CrossRef CAS PubMed .
49. L. M. Vanhee, H. J. Nelis and T. Coenye, Nat. Protoc., 2009, 4, 224–231 CrossRef CAS PubMed .
50. R. I. Amann, B. J. Binder, R. J. Olson, S. W. Chisholm, R. Devereux and D. A. Stahl, Appl. Environ. Microbiol., 1990, 56, 1919–1925 CrossRef CAS PubMed .
51. R. I. Amann, L. Krumholz and D. A. Stahl, J. Bacteriol., 1990, 172, 762–770 CrossRef CAS PubMed .
52. P. Belgrader, W. Benett, D. Hadley, J. Richards, P. Stratton, R. Mariella Jr. and F. Milanovich, Science, 1999, 284, 449–450 CrossRef CAS PubMed .
53. K. L. Josephson, C. P. Gerba and I. L. Pepper, Appl. Environ. Microbiol., 1993, 59, 3513–3515 CrossRef CAS PubMed .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nr05038b

‡ These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2023