⇦Chinese Physics Letters, 2022, Vol. 39, No. 5, Article code 058501 Highly Sensitive Mid-Infrared Photodetector Enabled by Plasmonic Hot Carriers in the First Atmospheric Window Yuan-Fang Yu (于远方)1, Ye Zhang (张也)2, Fan Zhong (仲帆)2, Lin Bai (白琳)1, Hui Liu (刘辉)2, Jun-Peng Lu (吕俊鹏)1*, and Zhen-Hua Ni (倪振华)1* Affiliations 1School of Physics, Southeast University, Nanjing 211189, China 2National Laboratory of Solid State Microstructures & School of Physics, Collaborative Innovation Center of Advanced Microstructures, Nanjing University, Nanjing 210093, China Received 3 March 2022; accepted 28 March 2022; published online 26 April 2022 *Corresponding author. Email: phyljp@seu.edu.cn; zhni@seu.edu.cn Citation Text: Yu Y F, Zhang Y, Zhong F et al. 2022 Chin. Phys. Lett. 39 058501    Abstract The first atmospheric window of 3–5 µm in the mid-infrared (MIR) spectral range pertains to crucial application fields, with particular scientific and technological importance. However, conventional narrow-bandgap semiconductors operating at this band, represented by mercury cadmium telluride and indium antimonide, suffer from limited specific detectivity at room temperature and hindered optoelectronic integration. In this study, a plasmonic hot electron-empowered MIR photodetector based on Al-doped ZnO (AZO)/bi-layer graphene heterostructure is demonstrated. Free electrons oscillate coherently in AZO disk arrays, resulting in strong localized surface plasmon resonance (LSPR) in the MIR region. The photoelectric conversion efficiency at 3–5 µm is significantly improved due to plasmon-induced hot-electron extraction and LSPR-enhanced light absorption. The specific detectivity reaches about $1.4 \times 10^{11}$ Jones and responsivity is up to 4712.3 A/W at wavelength of 3 µm at room temperature. The device's specific detectivity is among the highest performance of commercial state-of-the-art photodetectors and superior to most of the other 2D materials based photodetectors in the MIR region. These results demonstrate that a plasmonic heavily doped metal oxides/2D material heterostructure is a suitable architecture for constructing highly sensitive room-temperature MIR photodetectors.
cpl-39-5-058501-fig1.png
cpl-39-5-058501-fig2.png
cpl-39-5-058501-fig3.png
cpl-39-5-058501-fig4.png
 DOI:10.1088/0256-307X/39/5/058501 © 2022 Chinese Physics Society Article Text Mid-infrared (MIR) photodetectors for the first atmospheric window of 3–5 µm are highly significant for advanced applications, such as remote sensing,[1] communications,[2] and astronomy.[3] Conventional MIR devices based on epitaxially grown narrow-bandgap semiconductors, such as mercury cadmium telluride (MCT) and indium antimonide, still encounter challenges of limited detectivity at room temperature and hindered optoelectronic integration.[4,5] The high dark current of MCT photodetectors can result in increased noise, e.g., the generation–recombination noise makes it difficult to achieve high-performance MIR detection without a cryogenic environment. Moreover, complex cooling facilities severely hinder the optoelectronic integration of MCT in applications.[4] Graphene has recently attracted widespread interest in room-temperature photodetection due to its unique properties, such as high carrier mobility and broadband light absorption in ultraviolet-to-THz region.[6] Further, atomic thickness and the absence of dangling bonds make it easy for graphene to form van der Waals heterojunctions and on-chip optoelectronic integration.[7–9] However, intrinsic graphene photodetectors always suffer from low sensitivity, mainly due to weak light absorption[10] and short carrier lifetime.[11] Integrating plasmonic nanostructures into devices is an effective approach for photoelectric performance improvement, in which the induced near-field enhancement can prominently strengthen the light-harvesting capability of active materials.[12] In addition, localized surface plasmon resonances (LSPR) in nanostructures can damp nonradiatively by generating hot carriers via Landau damping.[13] The hot carrier harvesting before relaxation can prevent excess energy dissipation and carrier loss, resulting in efficient photoelectric conversion.[14] The issue of bandgap-limited spectral response range in infrared photon detectors can be addressed by hot carrier transfer.[15] Beyond noble metals, heavily doped metal oxides, such as Sn-doped In$_{2}$O$_{3}$ (ITO) and AZO, are excellent platforms for infrared plasmonic optoelectronics with low optical losses, which have composition-dependent LSPR ranging in near-infrared-to-MIR region.[16–18] In this study, we demonstrate a plasmonic hot carrier-enabled MIR photodetector based on AZO/bi-layer graphene heterostructure, which exhibits high detectivity at the first atmospheric window. Under MIR light illumination, free electrons at the conduction band of AZO are excited to a high energy level as energetic hot electrons, with LSPR damping. Ultrafast hot-electron transfer can overwhelm the competitive intraband cooling and prevent energy dissipation, improving the efficiency of infrared photoelectric conversion. Attributed to hot-electron transfer and improved light absorption, the AZO/bi-layer graphene hybrid devices exhibit a high detectivity of about $1.4 \times 10^{11}$ Jones and responsivity of about 4712.3 A/W at 3 µm at room temperature, which is superior to most 2D material-based MIR photodetectors, and nearly one order of magnitude higher than that in commercial MCT. These results demonstrate that constructing heterostructures with 2D materials and plasmonic heavily doped metal oxides are a suitable strategy for realizing on-chip integration of room-temperature MIR photodetectors. LSPR in AZO Arrays in MIR Region. To fabricate the AZO arrays, a poly(methyl methacrylate) (PMMA) layer was spin-coated on a SiO$_{2}$/Si substrate. Microhole arrays were patterned in the PMMA layer with standard lithography procedures. AZO (Al:ZnO = $2\!:\!98$) is sequentially deposited on the patterned PMMA film by thermal evaporation. Then, a lift-off process is implemented to remove PMMA with acetone, and the sample is annealed in a vacuum under 135 ℃ for an hour to improve the crystallinity.
cpl-39-5-058501-fig1.png
Fig. 1. LSPR in AZO arrays in the MIR region. (a) SEM image of prepared periodical AZO arrays with the thickness of about 100 nm (scale bar 5 µm). (b) EDS spectrum of AZO on the SiO$_{2}$/Si substrate. Insets: SEM and EDS maps of AZO arrays. The images with false colors correspond to the elements: Zn (green), O (red), Al (yellow), and Si (blue). (c) Extinction spectra of the annealed periodical AZO arrays with different diameters (1 to 4 µm) and the same edge-to-edge separation of 1 µm. (d) The peak position of LSPR as a function of diameter.
Figure 1(a) shows the scanning electron microscope (SEM) image of AZO micro-disk arrays on the SiO$_{2}$/Si substrate with a period of 4 µm and diameter of 3 µm. The thickness of the AZO micro-disk is about 100 nm (Fig. S1 in the Supporting Information). Energy-dispersive x-ray spectroscopy (EDS) and mapping were performed to identify the elemental composition of AZO arrays [Fig. 1(b)]. Al doping and the creation of oxygen vacancies can dramatically increase the free carrier concentration in AZO, thus increasing the Fermi level beyond the edge of the conduction band.[16] The coherent oscillation of free electrons in the AZO array supports a strong LSPR in the MIR region (approximately ranging from about 1.5 to 7 µm),[18] which is characterized using Fourier transform infrared spectroscopy [Fig. 1(c)]. As is expected, the resonance peak of the AZO disk shifts to longer wavelengths when the disk diameter increases, due to the reduced restoring force in the system [Fig. 1(d)], which is consistent with the behavior of LSPR-induced light absorption.[19] Photoelectric Conversion Mechanisms in AZO/bi-layer Graphene Heterostructure. Photodetectors based on AZO/bi-layer graphene heterostructure were fabricated [Fig. 2(a)]. Bi-layer graphene was mechanically exfoliated from highly oriented pyrolytic graphite and transferred onto a Si substrate with a 300-nm-thick SiO$_{2}$ capping layer. The Au electrodes (about 50 nm) with a Ni adhesion layer (about 5 nm) were fabricated via a standard lithography procedure using electron beam lithography, thermal evaporation, and lift-off procedures. Then, the AZO arrays are directly fabricated on the graphene channel. The output characteristic curve measured under dark conditions reveals the ohmic contact of the device, which ensures the effective collection of photogenerated carriers [Fig. 2(b)].[20] After the heterostructure formation, holes can transfer from AZO to graphene until an equilibrium state is reached, resulting in the p-doping of graphene. Under MIR illumination, free electrons at the conduction band of AZO can be excited to a high energy level as energetic hot electrons via LSPR damping,[21] and then enter the graphene channel, as depicted in Fig. 2(c). Ultrafast hot-electron transfer can overwhelm the competitive cooling process and prevent energy dissipation, thus improving the efficiency of infrared photoelectric conversion. Meanwhile, photogenerated holes are trapped in AZO with a certain spatial distribution, producing an additional electric field as a local gate voltage to modulate the channel conductance.[8] The devices have significant photoconductive gain due to the gating effects and hole trapping that prolong the carrier lifetime. The avoided energy dissipation by hot-electron transfer and the photogain mechanisms synchronously give the MIR AZO/bi-layer graphene photodetectors a high detectivity. Raman spectra of the bi-layer graphene and plasmonic AZO/bi-layer graphene heterostructure were obtained to verify the charge-doping scenario when forming heterostructures [Fig. 2(d)]. After forming heterostructures, the in-plane vibrational G-band of graphene blueshift from 1580 to 1582 cm$^{-1}$ (Fig. S2) indicates the occurrence of charge doping due to the nonadiabatic removal of the Kohn anomaly at the $\varGamma$ point. The blueshift of the 2D band from 2686 to 2690 cm$^{-1}$ after coupling with AZO (Fig. S3) demonstrates that graphene is doped by holes due to the charge-transfer-induced modification of the equilibrium lattice parameter.[22–24]
cpl-39-5-058501-fig2.png
Fig. 2. Photoelectric conversion mechanisms in the AZO/bi-layer graphene photodetector. (a) The schematic illustration of the device architecture. (b) The output characteristic curve measured under dark conditions for $V_{\rm g} = 0$ V. Inset at the bottom right is the optical image of the prepared device. (c) Schematic illustration of band diagrams, the excitation of plasmon-induced hot electrons under MIR laser excitation (yellow arrow), blocked interband transition under sub-bandgap excitation (yellow arrow with a black fork), hot-electron transfer (red arrow), the avoided carrier cooling (blue bending line), and intraband scattering in graphene (blue arc-shaped line). (d) Raman spectra acquired from AZO/bi-layer graphene heterostructure and pristine bi-layer graphene. Inset shows the Raman shift changes of the G and 2D bands of graphene before and after coupling with plasmonic AZO disk arrays. The wavelength of the excited laser is 532 nm, with a laser power of 400 µW.
Highly Sensitive Mid-Infrared Photodetection Enabled by Plasmonic Hot Carriers. The photoresponse of an AZO/bi-layer graphene hybrid photodetector under 3, 3.5, and 4 µm excitation is further characterized. The hybrid device, dominated by hot-electron transfer, shows dramatic photoresponse at room temperature [Fig. 3(a)]. Moreover, the response wavelengths ranging from 2.8 to 4 µm are consistent with the broadband LSPR profile at MIR [Fig. 3(b)]. We calculated the essential parameters of the AZO/bi-layer graphene photodetector to evaluate the device's performance. The detectivity $D^*$ can be calculated using the equation[25] $$ D^* =\frac{R\sqrt A }{\sqrt {2eI_{\rm dark} } },~~ \tag {1} $$ where $R$ is the responsivity, $A$ is the effective area of the device, $e$ is the electron charge, and $I_{\rm dark}$ is the dark current. Figure 3(c) shows that the maximum $D^*$ and responsivity of the AZO/bi-layer graphene device reach about $1.4 \times 10^{11}$ Jones and 4712.3 A/W at 3 µm, respectively. The device also exhibits comparable sensitivity at the wavelengths of 3.5 and 4 µm, which can be attributed to AZO's broad and robust MIR LSPR (Fig. S4). Under 3.5 µm illumination, the maximum $D^*$ and $R$ are as high as $4.2 \times 10^{10}$ Jones and 1481.7 A/W, respectively. Further, the device still maintains a high sensitivity under 4 µm illumination, with the maximum $D^*$ and $R$ of about $1.4 \times 10^{10}$ Jones and 477.9 A/W, respectively. We have also calculated the device's external quantum efficiency (EQE) using the equation[26] $$ {\rm EQE}=\frac{Rhc}{e\lambda },~~ \tag {2} $$ where $R$ is the responsivity, $e$ is the elemental charge, $h$ is the Planck constant, $c$ is the speed of light in vacuum, and $\lambda$ is the frequency of light. Figure S5 in the Supporting Information shows that the EQE of the device can reach about 194750%, 52490%, and 14810% at the wavelengths of 3, 3.5, and 4 µm, respectively, due to the effective hot-electron harvesting and photogain mechanisms. Notably, the high $D^*$, responsivity, and EQE can be further tuned by varying the ratio of the doped Al atoms, which significantly impacts the free carrier concentration and the MIR LSPR character. Another essential parameter of the photodetector is the response time. Curve fitting using the following double exponential functions extracts the rise time ($\tau_{1}$, $\tau_{2}$) and decay time ($\tau_{3}$, $\tau_{4}$) of the AZO/bi-layer graphene photodetector: $$ I_{\rm ph} =A_{1} \exp (-t/\tau_{1})+A_{2} \exp (-t/\tau_{2}),~~ \tag {3} $$ $$ I_{\rm ph} =A_{3} \exp (-t/\tau_{3})+A_{4} \exp (-t/\tau_{4}).~~ \tag {4} $$ According to the fitting results, the photoresponse of our devices at 3.5 µm is divided into two processes [Fig. 3(d)], in which $\tau_{1}$ is about 1.8 s and $\tau_{2}$ is about 13.5 s for the photocurrent increase. The decay time constant $\tau_{3}$ is about 3.1 s and $\tau_{4}$ is about 117.5 s. The relatively slow response is thought to be due to the defects trapping in photogenerated holes in AZO with shallow and deep states, which can extend carrier lifetimes and increase photoconductive gain.[8]
cpl-39-5-058501-fig3.png
Fig. 3. Photoresponse characterization of the AZO/bi-layer graphene MIR photodetectors at room temperature. $V_{\rm ds} = 0.5$ V and $V_{\rm g} = 0$ V. (a) Photoresponse of the device under different power densities at a wavelength of 3.5 µm. (b) Photocurrent of the devices as a function of wavelength at a power density of 480 mW/cm$^{2}$. (c) Power density-dependent detectivity and responsivity at a wavelength of 3 µm. (d) Transient response of AZO/bi-layer graphene photodetectors. The red and blue solid lines represent the fitting curves based on a double exponential function.
cpl-39-5-058501-fig4.png
Fig. 4. Performance comparison of the AZO/bi-layer graphene photodetectors with the previously reported photodetectors at room temperature. (a) Comparison of the specific detectivity between AZO/bi-layer graphene photodetector and other devices reported in the literature. Dash-dotted lines show the photovoltaic and photoconductive BLIP calculated for the field of view (FOV = 2$\pi$). (b) Comparison of responsivity of the AZO/bi-layer graphene photodetector with other reported MIR devices.
Figures 4(a) and 4(b) show the comparison of the AZO/bi-layer graphene photodetector with other MIR devices reported in literature. The detectivity and responsivity of our device are superior to most of the other advanced photodetectors based on low-dimensional materials,[3,27–31] such as b-AsP/MoS$_{2}$,[32] graphene/Ta$_{2}$O$_{5}$,[33] and B-doped Si/graphene.[34] Notably, the detectivity of the AZO/bi-layer graphene device at 3 µm is one order of magnitude higher than the existing commercial MCT detectors.[35] The highly sensitive, low-cost, and stable AZO/bi-layer graphene photodetector demonstrates significant potential in practical integrated applications, particularly for the first atmospheric window. It is emphasized that the performance of such hybrid photodetectors could be further improved by optimizing the interface and strengthening the hot carrier extraction. In addition, understanding the unrevealed hot carrier dynamics is particularly crucial for narrowing the gap between practical MIR systems and theoretical background-limited infrared performance (BLIP). In summary, we have demonstrated that coupling 2D materials with plasmonic heavily doped metal oxides is an effective approach to achieving highly sensitive MIR photodetectors in the first atmospheric windows of 3–5 µm. A plasmonic hot-electron-enabled MIR photodetector has been demonstrated on AZO/bi-layer graphene heterostructure. The device shows an ultrahigh detectivity of about $1.4 \times 10^{11}$ Jones and high responsivity of about 4712.3 A/W under 3 µm at room temperature. The specific detectivity of this device is comparable with that of commercial state-of-the-art photodetectors and higher than that based on 2D materials in the MIR region. The sensitivity of the photodetectors is expected to be improved further by facilitating hot carrier extraction and delaying hot carrier cooling using phonon bottleneck or Auger heating effects.[36,37] Acknowledgments. This work was supported by the National Key Research and Development Program of China (Grant No. 2017YFA0205700), the Strategic Priority Research Program of Chinese Academy of Sciences (Grant No. XDB30000000), the China Postdoctoral Science Foundation (Grant No. 2021M690625), and the Jiangsu Planned Projects for Postdoctoral Research Funds (Grant No. 2021K106B). References Phonon-enhanced photothermoelectric effect in SrTiO3 ultra-broadband photodetectorMulticolor Broadband and Fast Photodetector Based on InGaAs–Insulator–Graphene Hybrid HeterostructureFast Uncooled Mid‐Wavelength Infrared Photodetectors with Heterostructures of van der Waals on Epitaxial HgCdTeProgress, challenges, and opportunities for HgCdTe infrared materials and detectorsUltra-broadband graphene-InSb heterojunction photodetectorTwo-dimensional material nanophotonicsPhotodetecting and light-emitting devices based on two-dimensional materialsPhotogating in Low Dimensional PhotodetectorsMultispectral photodetectors based on 2D material/Cs3 Bi2 I9 heterostructures with high detectivityInvestigation of multilayer domains in large-scale CVD monolayer graphene by optical imagingPhotodetectors based on graphene, other two-dimensional materials and hybrid systemsShape-Controlled Synthesis and Surface Plasmonic Properties of Metallic NanostructuresPlasmon-induced hot-electron generation at nanoparticle/metal-oxide interfaces for photovoltaic and photocatalytic devicesPhotodetection based on Surface Plasmon-Induced Hot ElectronsHot Electron-Based Near-Infrared Photodetection Using Bilayer MoS2Transparent Conductive Oxides as Near-IR Plasmonic Materials: The Case of Al-Doped ZnO DerivativesPlasmonic behaviors of metallic AZO thin film and AZO nanodisk arrayUnconventional plasmonic sensitization of graphene in mid-infraredOptical properties of coupled metallic nanorods for field-enhanced spectroscopyChemical Vapor Deposition of Two-Dimensional PbS Nanoplates for PhotodetectionPlasmon-induced hot carrier science and technologyMonitoring dopants by Raman scattering in an electrochemically top-gated graphene transistorElectric Field Effect Tuning of Electron-Phonon Coupling in GrapheneRaman spectroscopy of graphene and graphite: Disorder, electron–phonon coupling, doping and nonadiabatic effects2D Metal Chalcogenides for IR PhotodetectionPhotocurrent generation with two-dimensional van der Waals semiconductorsA flexible, scalable, and self-powered mid-infrared detector based on transparent PEDOT: PSS/graphene compositeSpectrally Selective Mid-Wave Infrared Detection Using Fabry-Pérot Cavity Enhanced Black Phosphorus 2D PhotodiodesDirac plasmon-assisted asymmetric hot carrier generation for room-temperature infrared detectionNarrow bandgap oxide nanoparticles coupled with graphene for high performance mid-infrared photodetectionVan der Waals Epitaxial Growth of Mosaic‐Like 2D Platinum Ditelluride Layers for Room‐Temperature Mid‐Infrared Photodetection up to 10.6 µmRoom temperature high-detectivity mid-infrared photodetectors based on black arsenic phosphorusGraphene-Ta2 O5 heterostructure enabled high performance, deep-ultraviolet to mid-infrared photodetectionPlasmonic Silicon Quantum Dots Enabled High-Sensitivity Ultrabroadband Photodetection of Graphene-Based Hybrid PhototransistorsUnipolar barrier photodetectors based on van der Waals heterostructuresSlow Hot‐Carrier Cooling in Halide Perovskites: Prospects for Hot‐Carrier Solar CellsSlow cooling and highly efficient extraction of hot carriers in colloidal perovskite nanocrystals
[1] Lu X W, Jiang P, and Bao X H 2019 Nat. Commun. 10 138
[2] Cao G Q, Wang F, Peng M, Shao X M, Yang B, Hu W D, Li X, Chen J, Shan Y B, Wu P S, Hu L G, Liu R, Gong H M, Cong C X, and Qiu Z J 2020 Adv. Electron. Mater. 6 1901007
[3] Wang Y, Gu Y, Cui A L, Li Q, He T, Zhang K, Wang Z, Li Z P, Zhang Z H, Wu P S, Xie R Z, Wang F, Wang P, Shan C X, Li H, Ye Z H, Zhou P, and Hu W D 2022 Adv. Mater. 34 2107772
[4] Lei W, Antoszewski J, and Faraone L 2015 Appl. Phys. Rev. 2 041303
[5] Xu J B, Hu J X, Wang R B, Li Q, Li W W, Guo Y F, Liu F K, Ullah Z, Wen L, and Liu L W 2017 Appl. Phys. Lett. 111 051106
[6] Xia F N, Wang H, Xiao D, Dubey M, and Ramasubramaniam A 2014 Nat. Photon. 8 899
[7] Yu Y F, Miao F, He J, and Ni Z H 2017 Chin. Phys. B 26 036801
[8] Fang H H and Hu W D 2017 Adv. Sci. 4 1700323
[9] Zhang P, Zhang Y, Wang W H, Gao L, Li G F, Zhang S, Lu J P, Yu Y F, and Zhang J L 2021 Nanotechnology 32 415202
[10] Yu Y F, Li Z Z, Wang W H, Guo X T, Jiang J, Nan H Y, and Ni Z H 2017 J. Semicond. 38 033003
[11] Koppens F H, Mueller T, Avouris P, Ferrari A C, Vitiello M S, and Polini M 2014 Nat. Nanotechnol. 9 780
[12] Xia Y N and Halas N J 2005 MRS Bull. 30 338
[13] Clavero C 2014 Nat. Photon. 8 95
[14] Yu Y F and Ni Z H 2019 Laser & Optoelectron. Prog. 56 202403 (in Chinese)
[15] Wang W, Klots A, Prasai D, Yang Y, Bolotin K I, and Valentine J 2015 Nano Lett. 15 7440
[16] Calzolari A, Ruini A, and Catellani A 2014 ACS Photon. 1 703
[17] Lin J Y, Zhong K D, and Lee P T 2016 Opt. Express 24 5125
[18] Paria D, Vadakkumbatt V, Ravindra P, Avasthi S, and Ghosh A 2021 Nanotechnology 32 315202
[19] Aizpurua J, Bryant G W, Richter L J, de García A F J, Kelley B K, and Mallouk T 2005 Phys. Rev. B 71 235420
[20] Gu Y Y, Wang Y F, Xia J, and Meng X M 2020 Chin. Phys. Lett. 37 048101
[21] Brongersma M L, Halas N J, and Nordlander P 2015 Nat. Nanotechnol. 10 25
[22] Das A, Pisana S, Chakraborty B, Piscanec S, Saha S K, Waghmare U V, Novoselov K S, Krishnamurthy H R, Geim A K, Ferrari A C, and Sood A K 2008 Nat. Nanotechnol. 3 210
[23] Yan J, Zhang Y B, Kim P, and Pinczuk A 2007 Phys. Rev. Lett. 98 166802
[24] Ferrari A C 2007 Solid State Commun. 143 47
[25] Wang F K, Zhang Y, Gao Y, Luo P, Su J W, Han W, Liu K L, Li H Q, and Zhai T Y 2019 Small 15 1901347
[26] Buscema M, Island J O, Groenendijk D J, Blanter S I, Steele G A, van der Zant H S, and Castellanos-Gomez A 2015 Chem. Soc. Rev. 44 3691
[27] Zhang M and Yeow J T W 2020 Carbon 156 339
[28] Yan W, Shresha V R, Jeangros Q, Azar N S, Balendhran S, Ballif C, Crozier K, and Bullock J 2020 ACS Nano 14 13645
[29] Safaei A, Chandra S, Shabbir M W, Leuenberger M N, and Chanda D 2019 Nat. Commun. 10 3498
[30] Yu X, Li Y, Hu X, Zhang D, Tao Y, Liu Z, He Y, Haque M A, Liu Z, Wu T, and Wang Q J 2018 Nat. Commun. 9 4299
[31] Zeng L, Wu D, Jie J, Ren X, Hu X, Lau S P, Chai Y, and Tsang Y H 2020 Adv. Mater. 32 2004412
[32] Long M S, Gao A Y, Wang P, Xia H, Ott C, Pan C, Fu Y J, Liu E F, Chen X S, Lu W, Nilges T, Xu J B, XM W, Hu W D, and Miao F 2017 Sci. Adv. 3 e1700589
[33] Ho V X, Wang Y, Cooney M P, and Vinh N Q 2021 Nanoscale 13 10526
[34] Ni Z Y, Ma L L, Du S C, Xu Y, Yuan M, Fang H H, Wang Z, Xu M S, Li D S, Yang J Y, Hu W D, Pi X D, and Yang D R 2017 ACS Nano 11 9854
[35] Chen Y F, Wang Y, Wang Z, Gu Y, Ye Y, Chai X L, Ye J F, Chen Y, Xie R Z, Zhou Y, Hu Z G, Li Q, Zhang L L, Wang F, Wang P, Miao J S, Wang J L, Chen X S, Lu W, Zhou P, and Hu W D 2021 Nat. Electron. 4 357
[36] Li M J, Fu J H, Xu Q, and Sum T C 2019 Adv. Mater. 31 1802486
[37] Li M J, Bhaumik S, Goh T W, Kumar M S, Yantara N, Gratzel M, Mhaisalkar S, Mathews N, and Sum T C 2017 Nat. Commun. 8 14350