Long-wave infrared photothermoelectric detectors with resonant nanophotonics†

Yurong Zhang^a, Jiamin Jiang^a, Zhiheng Zhang^a, He Yu*^a, Yunlu Lian^a, Chao Han^a, Xianchao Liu^b, Jiayue Han^a, Hongxi Zhou^a, Xiang Dong*^a, Jun Gou*^a, Zhiming Wu^a and Jun Wang*^a
aState Key Laboratory of Electronic Thin Films and Integrated Devices, University of Electronic Science and Technology of China, Chengdu 610054, China. E-mail: wjun@uestc.edu.cn; yuhe@uestc.edu.cn; goujun@uestc.edu.cn; dongxiang@uestc.edu.cn
^bSouthwest Institute of Technical Physics, Sichuan, Chengdu 610041, China

First published on 14th August 2024

---

---

1. Introduction

Infrared (IR) machine vision stands as a transformative technology¹ that skillfully captures, converts, and processes a vast array of IR optical data from diverse objects. The demand² for long-wave and very long-wave infrared radiation detectors, covering wavelengths from 8 to 20 μm, is exceptionally high in critical sectors including remote sensing,³ thermal imaging,⁴ biomedical optics,⁵ molecular fingerprinting,⁶ medical imaging,⁷ and space communications.⁸ Traditionally, the detection of low-energy photons within these spectral ranges has relied on narrow-bandgap semiconductors, quantum wells, or heterostructures.^9,10 However, these materials generally require cryogenic cooling to achieve optimal performance, leading to systems that are not only bulky but also cost-prohibitive for many applications.¹¹ This challenge has sparked considerable interest in the development of room-temperature IR photodetectors that can be integrated on-chip, offering a promising solution to reduce both the size and weight of these devices, thereby enhancing their practicality and accessibility across various high-impact fields.¹²

Among the various types of IR detectors, thermal detectors stand out for their ultra-broadband spectral response,^13,14 which spans from near-infrared to terahertz, all operable at room temperature. These detectors measure temperature-dependent properties, such as resistance in bolometers,^15,16 temperature-difference-driven voltage in PTE detectors, and spontaneous polarization in pyroelectric (PE) detectors.¹⁷ Bolometers, often utilizing temperature-sensitive materials like vanadium oxide (VO_x)¹⁸ or amorphous silicon, typically require an external bias that introduces additional 1/f noise. On the other hand, PE detectors¹⁹ rely on variations in the spontaneous polarization vector with temperature changes, necessitating an external chopper to measure continuous-wave (CW) radiation power. PTE detectors,^20–23 operating through the Seebeck effect, excel with their ultra-broadband response that covers both IR and terahertz (THz) detection, without the necessity for cooling units, external power sources, or modulation choppers. The Seebeck effect in PTE detectors has been demonstrated across a diverse range of materials,²⁴ including carbon nanotubes,²⁵ III–V semiconductor nanowires,²⁶ and two-dimensional materials like graphene,²⁷ dichalcogenide materials,²⁸ and black phosphorus.²⁹ Despite their advantages, optimizing the photoresponse in these low-dimensional materials poses significant challenges, largely due to inadequate optical absorption and the early stages of material fabrication techniques.^30–34 Successfully addressing these challenges could significantly enhance the capabilities of PTE detectors, positioning them as fundamental components of next-generation IR imaging systems.³⁵ From another perspective, combining metamaterial structures with high absorption performance and photothermoelectric materials is an effective strategy for achieving high-performance PTE detectors. H. Atwater³⁶ designed sub-wavelength thermoelectric nanostructures with resonant spectrally selective absorption, capable of generating significant localized temperature gradients and photothermoelectric voltage. They demonstrated that such structure could achieve specific wavelength detection with an input power responsivity of up to 38 V W⁻¹ and a bandwidth close to 3 kHz. However, this has only been validated in the visible light spectrum. Extending this strategy to long-wave infrared still faces technical challenges in detector structure design and fabrication.

In this study, we introduce a high-performance PTE photodetector that combines a resonant nanophotonic structure with a thermoelectric nanofilm, achieving exceptional optical absorption and superior photothermoelectric properties. To enhance sensitivity in long-wave infrared (LWIR) detection, we have engineered a novel architecture incorporating a bismuth telluride (Bi₂Te₃) film within a photonic structure resonator, further optimized through plasmonic resonance to maximize absorption efficiency. Our comprehensive analysis, underpinned by both simulations and experimental validations, reveals that this innovative composite structure attains an average absorption rate exceeding 82.8% across the crucial 8–20 μm wavelength range, with peak absorption rates soaring up to 98.6%. The synergistic interplay between the broad absorption spectrum and the high Seebeck coefficient of the Bi₂Te₃ thin film endows the PTE detector with exceptional sensitivity and swift response to LWIR radiation at room temperature. We have also documented a substantial enhancement in the detector's response time, which can be attributed to the localized field enhancement effect provided by the metasurface structure. This effect significantly augments the efficiency of thermal carrier generation and transport. Moreover, the inclusion of a thin SiN_x layer as a protective coating ensures outstanding electrical and thermal stability, rendering the device well-suited for practical applications. This study delineates a promising avenue for the development of highly sensitive photodetectors and marks a significant leap forward in the field of room-temperature LWIR photodetection and imaging applications.

2. Results and discussion

The structural design and simulation results of the long-wave absorber are depicted in Fig. 1. This design includes a periodically arranged metallic metasurface, a metallic reflection layer, and a resonating cavity formed by a polyimide (PI) dielectric layer, as depicted in Fig. 1(a). Through meticulous optimization of geometric parameters, including the radius, period, and thickness of the metasurface, as well as the thickness of the resonating cavity, this structure achieves efficient absorption of long-wave infrared radiation. Titanium is employed for the metasurface, while aluminum serves as the material for the reflective layer. The periodic configuration of the metasurface is illustrated in Fig. 1(b). Fig. 1(c) illustrates the various resonance modes of the structure, each accompanied by their corresponding equivalent circuit diagrams, highlighting the dual resonant absorption modes typically found in metal–insulator–metal (MIM) metamaterial absorbers: propagating surface plasmon (PSP) resonance and localized surface plasmon (LSP) resonance.

Simulation studies were rigorously conducted to evaluate the optical absorption performance of the absorber structure. Incident light was directed along the Z-axis, covering an extended wavelength range from 8 to 23 μm. To ensure precise simulation results, periodic boundary conditions were applied along the X and Y axes, with the Z-axis bounded by perfectly matched layers. The primary goal of these simulation computations was to accurately determine the reflectance and transmittance of the structure at targeted wavelengths. The absorption coefficient, A(ω), was calculated based on the reflectance, R(ω), and transmittance, T(ω), providing a comprehensive understanding of the absorber's effectiveness across the specified wavelength range:

A(ω) = 1 − R(ω) − T(ω)	(1)

The radius, period, thickness of the metasurface, and height of the resonating cavity were deliberately chosen to explore the impact of various geometric parameters on the long-wave infrared absorption characteristics of the absorber, as illustrated in Fig. S1 (ESI†). The optimal structural parameters were determined through these simulations, resulting in specific geometric configurations: the periodic spacing (p) was set at 3.8 μm, radii of the larger and smaller disks (r and r₁) were optimized to 1.6 μm and 0.9 μm respectively, the metasurface layer thickness (t_top) was finalized at 12 nm, and the polyimide layer thickness (t_PI) was 1.5 μm. Under these conditions, the structure exhibited three prominent absorption peaks within the long-wave infrared spectrum, occurring at wavelengths of 9 μm, 13.3 μm, and 18 μm. Remarkably, these peaks achieved absorption rates of 98.6%, 98.5%, and 99.8% respectively, demonstrating the structure's high efficiency (Fig. 1d).

To further elucidate the coupled resonant absorption mechanisms of the absorber, detailed simulations were conducted to analyze the cross-sectional electromagnetic field distributions at three key resonant wavelengths: 9 μm, 13.3 μm, and 18 μm. As shown in Fig. 2(a), the electric field is predominantly concentrated within the top layer of titanium (Ti) discs. This localization is attributed to the excitation of surface plasmon polaritons (SPPs), which play a critical role in enhancing the absorptive properties of the structure. The magnetic field distribution analysis reveals two distinct resonant modes contributing to the absorption process. At the 9 μm wavelength, the absorption peak is driven by a combination of Fabry–Perot and PSP resonances within the dielectric layer. This coupling of modes effectively enhances the absorption at this specific wavelength. As the wavelength increases to 13.3 μm, PSP resonance becomes more pronounced and emerges as the predominant mechanism driving the main absorption peak. This intensification of the PSP resonance at mid-wavelengths underscores its significant role in the absorptive behavior of the structure. By 18 μm, the LSP resonance excited by the periodic array of discs predominates the absorption. The shift from PSP to LSP resonance as the wavelength increases highlights the dynamic interplay of resonant modes within the structure, each contributing differently across the spectrum.

In addition, the absorption behavior of the infrared metasurface to light of TM and TE mode at different angles was investigated. The absorption spectra for TE-polarized (φ = 0°) and TM-polarized (φ = 90°) waves at varying angles of incidence are depicted in Fig. 2(b). At smaller polarization angles, the absorptive structure exhibits an almost angle-independent absorption distribution for both TE and TM modes. As the angle of incidence increases, notable differences emerge in the absorption behavior between TE and TM polarized waves. The average absorption rate and bandwidth gradually decrease, primarily due to a reduction in the horizontal component of the incident magnetic field, which weakens the coupling and absorption strength. The disc-array-based absorptive structure is capable of exciting higher-order PSP resonance and LSP resonance modes at large angles of incidence within the long-wave infrared region. As shown in Fig. 2(c), the absorption characteristics of the structure remain largely unchanged with variations in the polarization angle of the incident light. This stability is attributed to the centrosymmetric design of the metasurface periodic structure, indicating an overall insensitivity of the structure to polarized light.

Based on the optimized design of the infrared absorber, experimental fabrication was conducted. A 200 nm thick aluminum metal film was deposited on a silicon wafer using magnetron sputtering technology. Subsequently, a polyimide (PI) dielectric layer approximately 1.5 μm thick was spin-coated on the aluminum surface to form a resonant cavity. A metasurface pattern was then photolithographed onto the PI surface, and a titanium (Ti) metal metasurface structure was fabricated, with geometric dimensions derived from the optimal simulation parameters. The cross-sectional and surface morphology of the experimentally prepared absorber is displayed in Fig. 3(a) and (b). Clear interfaces between the film layers can be seen in the cross-sectional scanning electron microscopy (SEM) images, and the thickness of each layer in the structure aligns well with expectations. Further analysis of the dimensions and thickness of the metasurface structure was performed using atomic force microscopy (AFM), as shown in Fig. 3(c) and (d). The AFM characterization indicated that the surface periodic structure was intact, with large and small disc diameters of approximately 2 μm and 1.2 μm, respectively, and a thickness of 54 nm.

The infrared absorption characteristics of the absorber were tested using a Fourier-transform infrared spectrometer (FTIR), with the absorption rate curves corresponding to different Ti metasurface thicknesses shown in Fig. 3(e). An increasing trend in absorption rate was observed with increasing metasurface thickness, reaching an average absorption rate of 82.8% in the 8–20 μm wavelength range. A comparison between the PI/Al structure and the Ti/PI/Al structure (Fig. 3f) revealed that absorption was primarily concentrated on the top layer of the titanium metasurface array, consistent with simulation results. Comparison between experimentally measured absorption rates and simulation results indicated close numerical agreement, though some discrepancies were noted; the three resonant peaks observed in simulations were not as distinct in experimental test results. X-ray photoelectron spectroscopy (XPS) characterization of the surface composition of the absorber revealed an unavoidable oxidation phenomenon on the titanium metal metasurface (Fig. S2, ESI†). Based on the experimental characterization results, the simulation model of the infrared absorber was revised by adding a 10 nm thick titanium oxide film above the titanium metasurface, as shown in Fig. S3 (ESI†). Recalculation of the absorptance using the revised simulation model showed good agreement between experimental and simulation results (Fig. S4, ESI†).

Leveraging the high absorption characteristics of the infrared absorber structure, a broad-spectrum detector utilizing the PTE mechanism was developed, as depicted in Fig. 4(a). The device preparation process is shown in Fig. S5 (ESI†). The sensitive material used is a piece of Bi₂Te₃ thin film, 200 nm thick, prepared via magnetron sputtering. The fabrication process involved sequential deposition of layers on a silicon substrate, beginning with the Si substrate, followed by a metal reflective layer, a PI resonant cavity, the bismuth telluride thermoelectric material, a SiN_x protective layer, and finally the metasurface structure. The material characterization of Bi₂Te₃ is shown in Fig. S6 (ESI†) and the Seebeck coefficient test is shown in Table S1 (ESI†).

The thickness of the PI is related to the temperature of the heating process as shown in Fig. S7 (ESI†). The silicon nitride layer served as an interface protective layer for the photolithographic patterning of the metasurface. Specific regions were reserved for electrode deposition during the preparation of the SiN_x protective layer to facilitate signal acquisition in the detector. A symmetrical electrode configuration was employed, achieving optimal PTE photocurrent through localized illumination of the device surface using a laser. This laser irradiation induced a temperature gradient across the device, resulting in a concentration gradient of thermal carriers and a photothermoelectric potential.

The I_ds–V_ds characteristics under dark conditions and illumination at 650 nm, 8 μm, and 10 μm are illustrated in Fig. 4(b). For the visible test in Fig. 4b, the power of the laser is 45 mW. For the longwave test in Fig. 4c, the laser power corresponding to 8 μm is 11 mW, The laser power of 10 μm is only 3.5 mW. The linear behavior of the I_ds–V_ds curve in the dark signifies nearly ohmic contacts between the electrodes and bismuth telluride, effectively mitigating the influence of the photovoltaic effect. Moreover, under visible and long-wave infrared illumination, the I_ds–V_ds curves exhibited downward shifts, indicative of photovoltage (V_ph) responses and photocurrent (I_ph) responses. Photocurrent measurements performed with a 0 V bias at 8 μm and 10 μm yielded photocurrent values of 24.7 nA and 15.8 nA, respectively, as depicted in Fig. 4(c).

The detector responsivity (R) and specific detectivity (D*) were calculated by the following equations:

	(2)

	(3)

	(4)

where I_light and I_dark represent the current values under illumination and in dark conditions, respectively, P_light is the incident light power density, A denotes the photosensitive area of the device, e is the elementary charge of an electron, Δf signifies the bandwidth, I_N is the noise current and NEP stands for the noise equivalent power. The noise current density of the device measured under 0 V and 1 V bias is illustrated in Fig. S8 (ESI†). Further evaluation of the detector's long-wave infrared response time is portrayed in Fig. 4(d) for the device under 8 μm laser irradiation. As observed, the response time of the detector in the long-wave infrared is in the millisecond range. The rise and fall times of the device under 8 μm illumination are 10 ms and 10 ms, respectively. For 10 μm illumination (Fig. S9, ESI†), the rise and fall times are 10 ms and 20 ms. Furthermore, a comparative study was conducted on the detector performance of Ti-Bi₂Te₃-PI-Al and Bi₂Te₃-PI optical structures. Both devices were subjected to 650nm laser irradiation, and the obtained I–t curves are shown in Fig. 4(e) and (f). The calculated results are depicted in Fig. 4(g), with the responsivity and D* of the device for long-wave 8μm is 0.388 mA W⁻¹ and 3.5 × 10⁷ jones, respectively. The test results indicate that the designed infrared absorber detector in this study achieved comprehensive improvements in sensitivity and response time compared to the reference devices. The enhancement in sensitivity is primarily attributed to the optimized design and increased absorptance. The corresponding improvement in response time is attributed to the localized field enhancement effect of the metasurface structure, which facilitates efficient generation and rapid transport of thermal carriers.³⁷

Further tests were conducted to evaluate the repeatability and stability of the detector. Benefiting from the SiN_x protective layer utilized in the metasurface photolithography process, the detector demonstrates outstanding air stability. Even after exposure to the atmosphere for 100 hours (Fig. 4h), there were virtually no alterations in the detector's photocurrent–time (I_ph–t) curves and response speed, underscoring the device's exceptional stability and repeatability. Following 50 cycles of pulsed illumination, the device's photoresponse remained at its initial value with no discernible degradation (Fig. 4i). We further perform the photocurrent mapping of the device as show in Fig. S10 (ESI†).

3. Conclusion

In this study, we propose a high-performance LWIR PTE photodetector based on resonant nano-photonics structures and thermoelectric nano-films. Through the synergistic interaction of metamaterial surfaces and photonic resonant cavities, an infrared absorber model spanning wavelengths from 8 to 20 μm was constructed using CST simulation. By optimizing structural parameters, broad-band perfect absorption was achieved. Utilizing the optimized absorber structure, LWIR absorbers spanning wavelengths from 8 to 20 μm was developed, exhibiting peak absorption and average absorption rates of 98.6% and 82.8%, respectively. By harnessing the high Seebeck coefficient of bismuth telluride thin films and the enhanced absorption of the perfect absorber, we have engineered a self-powered long-wave infrared PTE detector. This detector operates at room temperature and exhibits a responsivity of 0.388 mA W⁻¹, a response time of 10 ms, and exceptional air stability. By simply adjusting the resonant wavelengths of the absorbers, extension to other wavelengths can be easily achieved. This research approach provides a pathway for developing room-temperature, highly sensitive, and broad-spectrum infrared detectors.

4. Experimental section

4.1 Optical simulations

The numerical absorption spectrum and electromagnetic field distribution at absorption peaks of the absorber were theoretically simulated using CST simulation software. A single unit cell of the studied structure was simulated with periodic boundary conditions along the x and y axes, while perfect matching layers were employed along the direction of electromagnetic wave propagation (z-axis). In the simulation design of the circular disk array absorber structure, the dielectric constants of Ti and Al were obtained from the Drude model, with a PI dielectric layer chosen at a thickness of 1.5–2 μm and a dielectric constant set to 3.4. The thickness of the reflective layer was set to 200 nm, much greater than the skin depth of the incident wave in the metal, thus rendering its transmittance as 0. The thickness of the metasurface disks is ranged from 8 to 50 nm. Plane waves were emitted from the top surface of the unit cell, and the electromagnetic field distribution was illuminated using TM-polarized plane waves. For oblique incidence, the incident angle ranged from 0° to 70° with a step size of 5°. The structure is meshed locally, and the minimum accuracy is set to 1 × 10⁻⁴, and the total number of meshings exceeds 3 million, and the calculation results are relatively accurate.

4.2 Device fabrication

Initially, a 200 nm thick aluminum reflective layer was deposited on a silicon substrate using magnetron sputtering. Subsequently, a piece of PI was spin-coated (spin speed: 1000 rpm for 10 s; then 4000 rpm for 40 s) on the Al film. The PI was then cured by gradient heating. A 200 nm thick Bi₂Te₃ film was deposited on the dielectric layer by magnetron sputtering as the photo-thermoelectric material. Gold was then thermally evaporated on both sides of the device and used as electrodes. A silicon nitride (SiN_x) film was deposited on the surface of the dielectric layer with electrodes using magnetron sputtering. Subsequently, a photoresist was spun on the SiN_x prepared surface, and metasurface patterning was performed through pre-baking, contact exposure, post-baking, and development. Finally, a titanium film was deposited using magnetron sputtering, and periodic metasurface structures were obtained through a stripping process.

4.3 Device performance

The absorption spectrum was measured using a Spotlight200i Fourier transform infrared spectrometer. XPS measurements were performed using an ESCALAB XL system with a monochromatic Al X-ray source (1486.6 eV). SEM was employed to observe the surface morphology of the photoresist after development. AFM characterization was used to analyze the morphology and thickness of the metasurfaces. The I–V and I–t curves were measured using a PDA FS-Pro semiconductor analyzer with a probe station. Illumination in the visible/NIR range (405–1550 nm) was generated from light-emitting diodes, while the long-wave infrared photoresponse was measured using a quantum cascade laser. The rise and decay time (τ_on and τ_off) were defined as the time required for the net photocurrent to rise from 10 to 90% and decrease from 90 to 10% of its saturation value, respectively.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article [and/or its ESI†].

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was partially supported by National Natural Science Foundation of China (NSFC) (Grant No. 62175026, 62171094, 62305047), Project of the Sichuan provincial science and technology (Grant No. 2024NSFSC1444, 2024NSFSC0475), National Key Research and Development Program of China (Grant No. 2023YFB3611400), The China National Postdoctoral Program for Innovative Talents (Grant No. BX20230059), The China Postdoctoral Science Foundation (Grant No. 2023M740509), Aeronautical Science Foundation of China (Grant No. 20230024080001), and Guangdong Basic and Applied Basic Research Foundation, (Grant No. 2024A1515010005).

References

1. S. Fan and W. Li, Nat. Photonics, 2022, 16, 182–190 CrossRef CAS .
2. D. Li, X. Liu, W. Li, Z. Lin, B. Zhu, Z. Li, J. Li, B. Li, S. Fan and J. Xie, Nat. Nanotechnol., 2021, 16, 153–158 CrossRef CAS PubMed .
3. H. Zhu, Q. Li, C. Zheng, Y. Hong, Z. Xu, H. Wang, W. Shen, S. Kaur, P. Ghosh and M. Qiu, Light: Sci. Appl., 2020, 9, 60 CrossRef CAS PubMed .
4. N. Lee, T. Kim, J.-S. Lim, I. Chang and H. H. Cho, ACS Appl. Mater. Interfaces, 2019, 11, 21250–21257 CrossRef CAS .
5. W. Herschel, Philosoph. Magaz., 1800, 8, 9–15 CrossRef .
6. H. Qiu, Z. Yu, T. Zhao, Q. Zhang, M. Xu, P. Li, T. Li, W. Bao, Y. Chai and S. Chen, Sci. Chin. Informat. Sci., 2024, 67, 1–147 Search PubMed .
7. T. He, H. Ma, Z. Wang, Q. Li, S. Liu, S. Duan, T. Xu, J. Wang, H. Wu and F. Zhong, Nat. Photonics, 2024, 18, 60–67 CrossRef CAS .
8. R. Dash, S. Sahu, C. Mishra, K. Sethi, G. Palai and S. Sahu, Optik, 2016, 127, 9635–9639 CrossRef CAS .
9. R. A. Shelby, D. R. Smith and S. Schultz, Science, 2001, 292, 77–79 CrossRef CAS PubMed .
10. A. Tittl, A. K. U. Michel, M. Schäferling, X. Yin, B. Gholipour, L. Cui, M. Wuttig, T. Taubner, F. Neubrech and H. Giessen, Adv. Mater., 2015, 27, 4597–4603 CrossRef CAS .
11. W. Hu, X. Chen, Z. Ye and W. Lu, Appl. Phys. Lett., 2011, 99, 091101 CrossRef .
12. H. Cai, Y. Sun, X. Wang and S. Zhan, Opt. Express, 2020, 28, 15347–15359 CrossRef CAS .
13. F. Zhou, F. Qin, Z. Yi, W. Yao, Z. Liu, X. Wu and P. Wu, Phys. Chem. Chem. Phys., 2021, 23, 17041–17048 RSC .
14. Z. Qin, D. Meng, F. Yang, X. Shi, Z. Liang, H. Xu, D. R. Smith and Y. Liu, Opt. Express, 2021, 29, 20275–20285 CrossRef CAS .
15. K. Yu, P. Shen, W. Zhang, X. Xiong, J. Zhang and Y. Liu, Materials, 2022, 15, 2296 CrossRef CAS .
16. Z. Chen, H. Chen, H. Jile, D. Xu, Z. Yi, Y. Lei, X. Chen, Z. Zhou, S. Cai and G. Li, Diamond Relat. Mater., 2021, 115, 108374 CrossRef CAS .
17. W. Yang and Y.-S. Lin, Opt. Express, 2020, 28, 17620–17629 CrossRef .
18. Z. Ren, R. Liu, Y. Zhang, T. Liu, F. Li, X. Hong and Y. Guo, Mater. Lett., 2021, 296, 129832 CrossRef CAS .
19. N. Liu, M. Mesch, T. Weiss, M. Hentschel and H. Giessen, Nano Lett., 2010, 10, 2342–2348 CrossRef CAS PubMed .
20. F. Cheng, X. Yang and J. Gao, Opt. Lett., 2014, 39, 3185–3188 CrossRef PubMed .
21. Z. Yi, C. Liang, X. Chen, Z. Zhou, Y. Tang, X. Ye, Y. Yi, J. Wang and P. Wu, Micromachines, 2019, 10, 443 CrossRef .
22. T. D. Dao, S. Ishii, T. Yokoyama, T. Sawada, R. P. Sugavaneshwar, K. Chen, Y. Wada, T. Nabatame and T. Nagao, ACS Photonics, 2016, 3, 1271–1278 CrossRef CAS .
23. J. Y. Suen, K. Fan, J. Montoya, C. Bingham, V. Stenger, S. Sriram and W. J. Padilla, Optica, 2017, 4, 276–279 CrossRef CAS .
24. Y. Chen, Y. Wang, Z. Wang, Y. Gu, Y. Ye, X. Chai, J. Ye, Y. Chen, R. Xie and Y. Zhou, Nat. Electron., 2021, 4, 357–363 CrossRef CAS .
25. J. A. Montoya, Z.-B. Tian, S. Krishna and W. J. Padilla, Opt. Express, 2017, 25, 23343–23355 CrossRef CAS PubMed .
26. C. Wan, Y. Ho, S. Nunez-Sanchez, L. Chen, M. Lopez-Garcia, J. Pugh, B. Zhu, P. Selvaraj, T. Mallick and S. Senthilarasu, Nano Energy, 2016, 26, 392–397 CrossRef CAS .
27. F. Zhou, F. Qin, Z. Yi, W. Yao, Z. Liu, X. Wu and P. Wu, Phys. Chem. Chem. Phys., 2021, 23, 17041–17048 RSC .
28. S. Guenneau, C. Amra and D. Veynante, Opt. Express, 2012, 20, 8207–8218 CrossRef .
29. D. Schurig, J. Mock and D. Smith, Appl. Phys. Lett., 2006, 88, 041109 CrossRef .
30. S. Zhang, W. Fan, N. Panoiu, K. Malloy, R. Osgood and S. Brueck, Phys. Rev. Lett., 2005, 95, 137404 CrossRef .
31. J. B. Pendry, A. Holden, W. Stewart and I. Youngs, Phys. Rev. Lett., 1996, 76, 4773 CrossRef CAS PubMed .
32. R. A. Shelby, D. R. Smith and S. Schultz, Science, 2001, 292, 77–79 CrossRef CAS .
33. F. Ding, J. Dai, Y. Chen, J. Zhu, Y. Jin and S. I. Bozhevolnyi, Sci. Rep., 2016, 6, 39445 CrossRef CAS .
34. M. Luo, S. Shen, L. Zhou, S. Wu, Y. Zhou and L. Chen, Opt. Express, 2017, 25, 16715–16724 CrossRef CAS .
35. J. Hao, J. Wang, X. Liu, W. J. Padilla, L. Zhou and M. Qiu, Appl. Phys. Lett., 2010, 96, 251104 CrossRef .
36. K. W. Mauser, S. Kim, S. Mitrovic, D. Fleischman, R. Pala, K. C. Schwab and H. A. Atwater, Nat. Nanotechnol., 2017, 12, 770–775 CrossRef CAS PubMed .
37. J. Pettine, P. Padmanabhan, T. Shi, L. Gingras, L. McClintock, C.-C. Chang, K. W. Kwock, L. Yuan, Y. Huang and J. Nogan, Nature, 2024, 626, 984–989 CrossRef CAS PubMed .

---

Footnote

† Electronic supplementary information (ESI) available: Schematic parameters optimization of perfect absorber; XPS analysis of the metasurface; revised theoretical model according to the experimental result; comparison between simulated and experimental absorption characterization; device fabrication process schematic; AFM characterization of Bi2Te3 film; summary of seebeck coefficient of heat-sensitive materials; heating process and thickness analysis of the polyimide (PI) film; noise spectral of the device at different bias voltages; the time-resolved photocurrent of the device under 10 μm stimulation for analyzing its response speed; photocurrent mapping of the device. See DOI: https://doi.org/10.1039/d4tc02504k

This journal is © The Royal Society of Chemistry 2024