Promoting 2D Material Photodetectors by Optical Antennas beyond Noble Metals

The distinctive layered crystal structures and diverse properties of 2D layered materials (2DLMs) have established them as prospective building blocks for implementing next‐generation optoelectronics. One critical predicament in terms of light sensing is the weak absorption caused by the atomic‐scale thickness, as well as the limited effective wavelength range/low spectral selectivity constrained by the intrinsic band structures. Despite the fact that numerous noble metal antennas are harnessed for enhancing the light–matter coupling, they suffer from exorbitant cost and narrow resonant optical windows. To this end, a number of non‐noble plasmonic optical antennas have been developed to improve the light‐sensing properties of 2DLM photodetectors, and tremendous advances have been accomplished. Herein, a comprehensive overview of this subject is provided based on four aspects; namely, non‐noble metal antenna promoted 2DLM photodetectors, heteroatom doped semiconductor antenna promoted 2DLM photodetectors, non‐stoichiometric semiconductor antenna promoted 2DLM photodetectors, and MXene antenna promoted 2DLM photodetectors. The focus is on the device structures, preparation, and underlying mechanisms. In the end, the challenges are highlighted, and potential strategies addressing them are proposed, which aim to navigate the upcoming exploration in the related domains and fully exert the pivotal role of non‐noble plasmonic optical antennas toward advancing 2DLM photodetectors.


Introduction
Light sensing devices are the core components of modern information technology and intelligent systems, and they have DOI: 10.1002/adsr.202200079 been an important frontier of the current fundamental research. However, the limitations of photoelectric sensors based on traditional covalently bonded bulk materials (e.g., Si, In x Ga 1−x As, and Hg x Cd 1−x Te) have gradually emerged including large volume, high brittleness, rigorous operating conditions, as well as complicated preparation and processing. [1][2][3] Therefore, they can hardly adapt to the development trend of next-generation integrated optoelectronics. In this consideration, it is an impending demand to explore new light-sensing material systems as potential alternatives. 2D layered materials (2DLMs) are burgeoning photoelectric elements with distinctive layered crystal structures. [4][5][6][7][8] They have manifested a series of advantages including naturally-passivated surface, thickness/strain/torsion-regulated bandgap, excellent in-plane carrier mobility, Si-complementary metal-oxidesemiconductor processing compatibility, outstanding flexibility, etc. Thus far, hundreds of 2DLMs have been explored including elemental semiconductors and their derivatives, [9][10][11][12][13] nitrides, [14] phosphides, [15,16] transition metal dichalcogenides, [17][18][19][20][21][22][23][24][25][26][27][28] post transition metal chalcogenides, [29][30][31][32][33][34][35] transition metal halides, [36][37][38] solid solutions, [39] multi-element compounds, [40][41][42][43][44][45][46][47][48][49][50][51][52][53][54][55][56] topological insulators, [57,58] alloys, [59] etc. Their bandgap values range from 0 up to 6 eV, theoretically enabling them to meet the diverse practical applications in various wavebands. Taking advantage of the above merits, 2DLMs have been regarded as a transformative candidate platform for the implementation of the next generation of optoelectronic devices, attracting worldwide research attention in both academic and industrial communities. [4,6,[60][61][62][63][64][65][66] In recent years, researchers have made numerous accomplishments in the field of 2DLM optoelectronic devices. However, some 2DLM photodetectors suffer from poor responsivity below 10 mA W −1 , [67,68] which is predominantly due to the weak light absorption caused by their atomic-scale thickness. As proof, the absolute light absorption of monolayer graphene and transition metal dichalcogenides is only in the range of ≈2-10%. [69,70] This has severely restricted the accumulation of photogenerated carriers, thus being a primary obstacle impeding the further improvement of the performance of the corresponding photodetectors. To address the above issue, researchers have developed a plethora of device optimization strategies. For example, by coupling Au plasmonic nanostructure arrays, Miao et al. have improved the photoresponse of a few-layer MoS 2 photodetector by ≈200%. [71] In another study, Bang and collaborators have demonstrated the improvement of the responsivity of monolayer MoS 2 photodetectors by ≈2-3 orders of magnitude by coupling Ag plasmonic nanowires. [72] Basically, the localized surface plasmon resonance (LSPR) of noble metal nanostructures, that is, the confined collective oscillation of free electrons gas with the concordant frequency as incident irradiation in nanoscale, leads to enhanced near field and thus increased light absorption by concentrating the light energy from the free space into subwavelength volumes, which, in turn, excites a large number of non-equilibrium carriers contributing to the improved photosensitivity. On the other hand, the performance of 2DLM photodetectors can also be improved through the integration of an optical cavity or optical waveguide. As an example, in 2020, Flöry et al. achieved both high sensitivity and fast response rate (0.2 A W −1 , 50 GHz) based on a MoTe 2 /graphene van der Waals heterojunction photodetector integrated with a Si optical waveguide. [73] Although substantial progress has been made, the preceding studies suffer from their own shortcomings, which have seriously hindered their further commercialization. On one hand, precious metal-based optical antennas have the deficiencies of high material cost and narrow resonance wavelength range. On the other hand, the integration of an optical cavity or optical waveguide with 2DLMs is demanding and subjected to complicated microfabrication procedures. [73,74] By this token, it is still quite essential to develop cost-efficient optical antennas with wide-range resonance spectrum and easy integration toward ameliorating 2DLM photodetectors.
Profited from the numerous advantages, non-noble material plasmonics has garnered worldwide research enthusiasm in the past few years. [75] Since these materials do not contain precious elements, non-noble optical antennas have exhibited huge advantages in terms of fabrication cost as compared to the widely explored precious metal ones (such as Au, Ag, Pt, and Pd). In addition, non-noble metal optical antennas have exhibited resonance frequency across a substantially broad spectral range across ultraviolet (UV) to far-infrared (FIR), [76][77][78][79] which far exceeds that of the classical noble metal antennas (normally limited to the UV to  [94,97,111] Copyright 2015, 2017, 2022, American Chemical Society. Reproduced with permission. [100] Copyright 2019, John Wiley and Sons. visible (vis) range due to high charge carrier concentration on the order of 10 22 -10 23 cm −3 ). [80] This enables the customization of optical antennas with highly matched effective optical windows to various 2DLMs manifesting diverse band structures (bandgap ranging from 0 to 6 eV). [4,[81][82][83] Moreover, the categories of nonnoble metal optical antennas are much more diverse, and their electronic band structures (i.e., conduction band minimum, valence band maximum, bandgap, Fermi level, etc.) are thus much more abundant than metals. This has provided a rich library for the design of various antenna/semiconductor heterointerfaces, which imparts an additional degree of freedom for the regulation of the dynamic process of photocarriers that is critical to hot carrier extraction. Thus far, a number of competitive non-noble plasmonic competitors have emerged including Al, [84] Cu, [85] Si, [86] Ge, [87] TiN, [88] Cu x S, [89] Cu 2-x Se, [90] MoO 3-x , [91] WO 3-x , [92] MXene, [93] and so on.
Over the past decade, researchers have developed a host of nonprecious optical antennas toward improving the performance of 2DLM photodetectors. [94][95][96][97][98][99][100][101][102][103][104][105][106][107][108][109][110][111] However, these studies are scattered. In addition, to our knowledge, most of the literature reviews regarding the optical engineering of 2DLM photodetectors have been focused on precious metal optical antennas. [11,23,[112][113][114] There still lacks a comprehensive overview on improving 2DLM photodetectors with non-precious optical antennas. In this review, the advances in non-precious plasmonic optical antennapromoted 2DLM photodetectors, including non-noble metals, doped semiconductors, non-stoichiometric semiconductors, and MXenes (Figure 1), [94,97,100,111] have been systematically summarized to provide an explicit panoramagram. In the end, the critical challenges encountered in this domain are put forward and potential solutions addressing them are proposed, aiming at navigating future endeavors.  ) and an InSe photodetector coupled with Al nanodisks (red line). The green dash line is the scattering spectrum of the Al nanodisks. c) Channel current as a function of bias voltage of a pristine InSe photodetector (blue line) and an InSe photodetector coupled with Al nanodisks (red line) under illumination. The black line is the dark current. Reproduced with permission. [94] Copyright 2015, American Chemical Society. d) 3D schematic diagram of an h-BN photodetector coupled with Al NPs. e) Channel current as a function of the power density of a pristine h-BN photodetector (green line) and h-BN photodetectors coupled with S 20 (average particle size of ≈20 nm) and S 30 (average particle size of ≈30 nm) Al NPs (blue and red lines) upon 205 nm illumination. f) 3D distribution of the electric field intensity under light excitation with a wavelength of 210 nm in Al NPs of 20 nm size with an interparticle separation of 10 nm. Reproduced with permission. [135] Copyright 2022, American Chemical Society.
As a pioneering attempt, in 2015, Lei et al. demonstrated the improvement of photosensitivity of an InSe photodetector by capitalizing non-precious Al nanodisks as the optical antennas (Figure 2a). [94] Periodic Al nanodisks are patterned onto the InSe channel by e-beam lithography and e-beam evaporation. As shown in Figure 2b,c, the photoresponse of the InSe photodetector increases markedly across the UV to near-infrared (NIR) range with the modification of Al nanodisks. Importantly, the scattering peak of Al nanodisks is largely consistent with the peak of the photoresponse spectrum of the InSe photodetector coupled with Al nanodisks (≈510 nm), suggesting that the fundamental origin for the promoted performance is the enhanced light absorption induced by the plasmonic resonance of the Al antennas.
Despite that photolithography and electron beam lithography are currently the most precise and repeatable approaches to integrating metal nanostructures with customized sizes and shapes, these techniques are relatively cost-inefficient and timeconsuming. Therefore, it has always been a worthy research subject to explore alternative techniques for preparing plasmonic metal nanostructures in consideration of squeezing costs. To this end, in 2022, Kaushik et al. demonstrated the improvement of h-BN deep-UV photodetectors by coupling them with selfassembled Al nanoparticle (NP) optical antennas (Figure 2d). [135] In this study, a facial and cost-efficient thin-film dewetting strategy has been developed for preparing dispersive Al NPs. Basically, Al nanofilms are first deposited onto h-BN samples using electron-beam evaporation, followed by rapid thermal annealing at 600°C for 150 s in an inert nitrogen atmosphere. As shown in Figure 2e, with the modification of Al NPs with the size of 20/30 nm, the photocurrent upon 205 nm irradiation has been elevated by ≈5.5/26 times, while the response rate remains largely unaffected, making this strategy outperform the widely explored quantum dot (QD) modification approach to some www.advancedsciencenews.com www.advsensorres.com extent, [113,136,137] where the response speed is inevitably sacrificed due to the severe carrier trapping induced prolonging of the decay process. To have an in-depth insight, the 3D electric field intensity distribution of the Al NPs is extracted by performing a systematic numerical simulation ( Figure 2f). Evidently, the occurrence of intense LSPR in Al NPs upon UV excitation (210 nm) is demonstrated, which efficiently squeezes the free-space light into sub-diffraction volume and thereby increases the light absorption of the underlying h-BN layer. In addition, the coupling among the Al NPs leads to enhanced local electric field intensity in the substrate, contributing to the expedited spacial separation of photogenerated electron-hole pairs. Furthermore, the local electric field intensity can be flexibly engineered by simply modulating the size of the Al NPs, providing an additional degree of freedom for tuning the device properties. Of note, the enhancement factor of this study outperforms those of classical noble metal optical antenna-promoted 2DLM photodetectors in the UV range, [138] demonstrating outstanding competitiveness in preparing the next generation of short-wave 2DLM photodetectors. This is because the bulk plasmon frequency ( p ) of a plasmonic active material is directly related to its free carrier density (n) where e is the elementary charge, 0 is the dielectric permittivity of vacuum, and m* is the effective mass of free carriers. As for nanoparticles in the quasi-static limit (i.e., small enough to enable instantaneous electromagnetic interaction between different parts and the excitations are limited to dipolar modes), the LSPR will result in pronounced light absorption under the following resonance condition where LSPR is the resonance frequency of the collective plasma oscillation, m is the dielectric constant of the surrounding medium, and is related to the carrier damping. Since the number of valance electrons of Al is more than that of Au/Ag, Al exhibits a higher bulk plasma frequency (≈15 eV) than these conventional noble metals (≈8-9 eV). [139] As a consequence, the plasmonic resonant peak of Al antennas commonly occurs at shorter wavelengths.
In addition to being adapted in the photoconductive-type photodetectors, non-noble metal plasmonic optical antennas have also been developed for ameliorating the 2DLM-based heterojunction photodetectors. In 2017, propelled by the plasmonic Cu NPs, a highly sensitive graphene/CdSe vertical heterojunction photodetector in the red and infrared (IR) spectral range is demonstrated by Wang et al. [96] In this study, hexagonal Cu NPs are functionalized onto the graphene layer by a polystyrene nanosphere-assisted patterning approach (Figure 3a,b). Importantly, systematic measurements determine that the photoresponse of the Cu NPs@graphene/CdSe heterojunction photodetector is dozens of times that of the pristine graphene/CdSe device upon light illumination across the red and IR spectral range ( Figure 3c). Finite element simulation reveals pronounced light absorption of Cu NPs in the range of 700 to 900 nm (Figure 3d), revealing an intense LSPR effect that accounts for the boosted photosensitivity. This has also been demonstrated by the electric field distribution of the Cu NP array under light irradiation with wavelengths of 780 and 900 nm (Figure 3e,f). In principle, energetic non-equilibrium hot electrons with a high kinetic energy of ≈2 eV [140,141] (modulated by the size, shape, and surrounding dielectric medium) are generated in Cu NPs upon resonant light excitation followed by non-radiative electromagnetic decay of the surface plasmons, which subsequently escape from the Cu NPs and inject into the underlying graphene/CdSe heterojunction ( Figure 3g). As a consequence, these non-equilibrium free carriers are swept by the interfacial built-in electric field and then collected by the external circuit, thus contributing to the improved photocurrent.
On account of the relatively large interlayer distance, nonprecious metals can also be integrated within the van der Waals gaps of 2DLMs. [142] For example, Han et al. have demonstrated the formation of intercalated Cu carpets within MoS 2 via a vacancy-mediated atomic diffusion strategy. [143] Up to now, a myriad of approaches have been developed for the atomic intercalation of various non-noble metals spanning Fe, [144] Co, [145] Sn, [146,147] Cu, [148] etc. In this respect, it is theoretically feasible to integrate non-noble metal optical antennas within 2DLMs through atomic intercalation techniques to improve their interaction with the incident light. To this end, in 2021, Stern et al. demonstrated enhanced optoelectronic activity of MoS 2 via copper intercalation by a wet-chemical process (Figure 4ac). [36] Of note, the Cu-MoS 2 /Si heterojunction photodetector exhibits much-improved photocurrent, responsivity, and photogain as compared to a pristine MoS 2 /Si counterpart device across a broad spectrum range of ≈500-1100 nm (Figure 4d-f). Specifically, the maximum responsivity of the Cu-MoS 2 /Si heterojunction photodetector reaches 4.2 × 10 4 A W −1 , which is 5.1 times that of a pristine MoS 2 /Si device. In addition, this value is also higher than most previously reported devices with similar device architectures. [149][150][151] In theory, atomic intercalation can isolate the inserted metals from the ambient environment to a certain extent. In this respect, compared to the surface modification scheme, the atomic intercalation strategy is much more conducive to the long-term stability of the optical antennas, which is of great significance for easily oxidized plasmonic metals such as Al and Ag. However, atomic intercalation will inevitably induce defects, such as lattice distortion, that may degrade the transport properties of the targeted 2DLMs, which also needs to be systematically taken into account. In addition, how to control the size and morphology of the intercalated atomic clusters, which play a critical role in the plasmonic properties, is also a critical issue to be addressed in the future.

Heteroatom Doped Semiconductor Optical Antenna Promoted 2DLM Photodetectors
As is well known, LSPR represents a physical phenomenon of the coherent and collective oscillation of free electron gas in micro-/nano-scale solid structures upon the stimulation of resonant incident light. [152,153] That is, the excitation of LSPR is closely associated with the interactions between incident light Figure 3. a) 3D schematic illustration of the fabrication procedures for constructing Cu NPs array. b) 3D schematic diagram of a graphene/CdSe Schottky junction photodetector coupled with Cu NP antennas. c) Responsivity as a function of wavelength of the graphene/CdSe (black line) and Cu NPs@graphene/CdSe (blue line) heterojunction photodetectors. d) Experimental absorption spectrum (black line) and theoretical absorption spectrum from finite element simulation (red line) of the Cu NPs. e,f) Electric field intensity distribution of the Cu NP antennas under illumination with wavelengths of 780 and 900 nm, respectively. g) Energy band diagram illustrating the working mechanism of the Cu NPs@graphene/CdSe heterojunction photodetector. Reproduced with permission. [96] Copyright 2017, Royal Society of Chemistry. and free carriers. By this token, in theory, non-metal semiconductor nanostructures can thus also host the potential to exhibit pronounced LSPR effect through controlled heteroatom doping to induce a targeted density of free carriers supporting collective oscillation on account of the discrepancy in the number of valance electrons between the enthetic atoms and the host atoms (i.e., the substituted atoms). For example, in 2018, Zandi et al. revealed that the resonant peak of the LSPR extinction spectrum of Sn:In 2 O 3 (STO) nanocrystals (NCs) markedly shifted from ≈3000 to ≈5500 cm −1 with the nominal Sn doping level ranging from 1% to 10%. [154] As a consequence, non-metal heteroatomdoped semiconductor plasmonic nanostructures have garnered extensive research enthusiasm in recent years in the context of their advantage in large-scale applications enabled by their competitive material cost. Thus far, a vast number of highly compelling doped semiconductor-based plasmonic nanomaterials have emerged such as indium tin oxide (ITO), [155] STO, [154] Tedoped Si, [78] H-doped MoO 3 , [156] F and In co-doped CdO, [157] Aldoped ZnO, [158] H-doped MoO 3 , [159] Ga 2 FeO 4 , [160] Cu/In-doped Cd 2 SnO 4 , [161] and so on. Compared with classical metal optical antennas, semiconductor-based optical antennas have exhibited distinct advantages. On one hand, the doping concentration of semiconductors, which plays a critical role in the bulk plasmonic resonance frequency, can be more precisely and widely regulated (across several orders of magnitude) via controlled doping, [159] whereas the free carrier concentrations of metals are high and relatively unmodifiable (≈10 22 -10 23 cm −3 ). [80] This theoretically endows the semiconductor-based plasmonic optical antennas with a broadly regulated resonance spectrum that ranges far beyond the scope of metallic ones, [162,163] making them essential complements to the metal antennas. In addition, the intrinsic interband excitation of these plasmonic active semiconductors provides an additional pathway to harvest the light energy. On the other hand, the categories of semiconductors are much more diverse than metals, [75,164] and their electronic band structures (i.e., conduction band minimum, valence band maximum, bandgap, Fermi level, etc.) are thus much more abundant. This has provided a substantially rich library for the design of various antenna/2DLM heterointerfaces, which imparts an additional degree of freedom for tuning the dynamics of photoexcited hot carriers. In this regard, the integration of heteroatom-doped semiconductor optical antennas with 2DLMs with a broad distribution of bandgap values to boost their photosensitivity and broaden their functionalities is an appealing research subject of interest in the optoelectronic industry.
Unlike conventional metal nanostructures normally exhibiting LSPR bands in the vis and UV regions, [121,[165][166][167][168] the heteroatom-doped semiconductor optical antennas typically manifest LSPR in the IR spectrum, which thereby well complement the classical metal optical antennas. In this respect, these materials host indisputable potential for broadband light sensing and pertain to crucial applications such as remote sensing, military surveillance, and astronomy. For a proof of concept, in 2017, Ni et al. developed Si QDs heavily doped by B for the modification of graphene photodetectors toward implementing high-performance broadband light sensing (Figure 5a). [97] Silicon hosts extensive advantages spanning earth abundance, high air-stability, and nontoxicity. The B-doped Si QDs are synthesized by a non-thermal plasma method, where SiH 4 and B 2 H 6 are exploited as the precursors. Absorption measurement determines that a prominent LSPR peak of B-doped Si QDs occurs at ≈3 m, which can be well explained by the Mie absorption theory including the Drude contribution (Figure 5b). Finite-difference timedomain (FDTD) simulations conclude that the LSPR of B-doped Si QDs will induce an intense electromagnetic field in their vicinity (i.e., near-field enhancement) upon 3 m resonant illumination ( Figure 5c). Of note, the enhanced electromagnetic field is conducive to increasing the light absorption (A) of the underlying graphene layer according to [169] A ∝ where |E 0 | is the electric amplitude of incident light, |E P | is the electric amplitude in the presence of a plasmonic effect, and is the fine structure constant. Therefore, more photocarriers can be generated with the modification of B-doped Si QDs, thus resulting in a boosted mid-infrared (MIR) photoresponse. Benefiting from the pronounced LSPR effect in the IR spectrum range, the hybrid B-doped Si QDs/graphene photodetectors demonstrate competitive photosensitivity across the vis to MIR range (Figure 5d,e). Specifically, the responsivity reaches ≈10 A W −1 in the MIR region, whereas the pristine Si QDs/graphene photodetector exhibits negligible photoresponse in this waveband on account of the weak light absorption. This responsivity value far outperforms most of the reported 2DLM-based MIR photodetectors. [170][171][172] In general, this study puts forward a unique protocol for implementing high-performance long-wave 2DLM photodetectors by synergistically capitalizing on the intense LSPR effect of doped semiconductor optical antennas, as well as the high mobility of 2DLMs. In a subsequent study, Yu et al. realized a high-performance room-temperature bi-layer graphene MIR photodetector by decorating the 2D channel with Al-doped ZnO disk arrays. [110] Remarkably, upon 3 m illumination, the responsivity and detectivity of the device reach 4712.3 A W −1 and 4 × 10 11 Jones, respectively. The outstanding photosensitivity is associated with the synergy of plasmoninduced hot-electron injection and LSPR-enhanced light absorption.
Apart from the photoconductive-type photodetectors, heteroatom-doped semiconductor optical antennas have also been exploited for ameliorating 2DLM-based photodiodes. As an example, Lu et al. have demonstrated improved NIR photosensitivity of the graphene/Ge vertical heterojunction photodetectors by decorating the graphene layer with ITO NP optical antennas (Figure 6a). [95] The ITO NPs are prepared via a reflux method, in which the doping level of Sn can be flexibly controlled by tuning the In/Sn precursor ratio. As shown in Figure 6b, as the Sn content increases from 5% to 30%, the LSPR absorption peak of the ITO NPs first blueshifts from ≈1730 to ≈1580 nm, and then redshifts from 1580 to 1760 nm. The strong LSPR effect in the IR spectrum has also been consolidated by FDTD simulations, which reveal that the LSPR can be further tailored by the size of the ITO NPs as well (Figure 6c). These have provided a wide range of degrees of freedom for tailoring the optical properties to match various 2DLMs and working conditions. Systematic transport measurements determine that the ITO NPs/graphene/Ge heterojunction photodetector demonstrates a high responsivity of 185 mA W −1 , an excellent detectivity of 2.28 × 10 13 Jones, and an ultrashort response/recovery time of 450/460 ns upon 1550 nm illumination under a self-driven mode, all of which far outperform those of a pristine graphene/Ge photodetector. The remarkable performance improvement is ascribed to the strong light-confinement capacity of the ITO NP antennas, which results in enhanced light-matter interactions as well as high-efficient injection of hot electrons (Figure 6d,e). Figure 6. a) 3D schematic diagram of an ITO NPs@graphene/Ge heterojunction photodetector. b) Absorption spectra of ITO NPs with different doping levels (5%, 10%, 20%, 30%). c) The electric field intensity distribution of ITO NPs (10% Sn) with different sizes (30,24,20, and 14 nm). d) Electric-field intensity distribution of Ge, graphene/Ge, and ITO NPs@graphene/Ge. e) Absorption spectra of various structures. Reproduced with permission. [95] Copyright 2016, John Wiley and Sons.
With the rapid expansion of research enthusiasm and progressive production techniques, 2DLMs have grown into a vast library with numerous members up to several hundred by far, which host a wide variety of electronic band structures. It gives rise to their extremely abundant optoelectronic properties, which can theoretically cover the multi-scene practical demands of optoelectronic devices in various wavebands. [4] As proof, in a previous study, Yao et al. demonstrated roomtemperature terahertz light sensing by exploiting a topological insulator Bi 2 Te 3 /Si heterojunction photodetector. [57] In another study, Wu et al. reported on the UV to FIR light sensing based on a PdSe 2 /CdTe photodiode. [173] However, the photosensitivity of the above-mentioned prototype devices in the long-wave spectral region is still relatively low, which is insufficient to satisfy the requirements of practical applications. Thus far, although preliminary progress on improving 2DLM IR photodetectors by heteroatom doped semiconductor plasmonic optical antennas has been achieved, there is still a severe lack of related research in the long-wave infrared and terahertz optical wavebands useful for space remote sensing, military reconnaissance, non-invasive imaging, and medical examination. In reality, a host of doped semiconductor optical antennas with pronounced plasmonic effects in the long-wave spectrum region have been developed in the past few years. For example, in 2017, Omeis et al. demonstrated that the Si-doped InAsSb-based metal-insulator-metal antenna array can support gap-plasmon resonances up to terahertz frequencies (≈7 THz). [174] In 2022, Wang et al. unveiled that MIR LSPR can occur in Si films hyperdoped with deep-level impurity tellurium, and such LSPR can be further enhanced and spectrally extended to the FIR spectral range by fabricating 2D arrays of micrometer-sized antennas. [78] On basis of the above inspiring advancements, coupling these long-wave non-noble plasmonic optical antennas with the long-wave sensitive 2DLMs definitely represents an attractive research topic in the upcoming future.

Non-Stoichiometric Semiconductor Optical Antenna Promoted 2DLM Photodetectors
Apart from heteroatom doping exerting polyvalent state combination, free carriers can also be intentionally introduced to semiconductors by finely designing their atomic stoichiometric ratio. Therefore, in theory, LSPR can also be artificially induced in non-stoichiometric semiconductor NCs. To date, a plethora of approaches, such as metal reduction methods, [20,44] hydrogen reduction, [175] and plasma exposure, [176] have been developed to prepare non-stoichiometric semiconductors. Importantly, it is widely revealed that non-stoichiometric semiconductor NCs have exhibited pronounced LSPR. As a consequence, these materials can benefit from outstanding light absorption enhancement in the spectral range beyond their intrinsic bandgap limitations. For example, in 2019, Gong et al. demonstrated a controlled extension of the absorption spectral range of ion-deficient Fe 1-x S 2 NCs from vis to short-wave infrared (1-3 m) by slightly modulating the x content from 0.01 to 0.107. [21] In recent years, non-stoichiometric semiconductor NCs have been extensively investigated for numerous realms spanning photocatalysis, [177] surface-enhanced Raman scattering, [178] sustainable cancer therapy, [179] and photoelectrocatalysis, [37] on account of a variety of advantages such as low preparation cost, as well as a well controllable composition, shape, and size. Thus far, a large number of non-stoichiometric semiconductor optical antennas with intense LSPR effect have emerged, such as MoO 3-x , [44] Cu 3-x P, [61] Cu 2-x S, [62] Cu 2-x Se, [180] WO 3−x , [63] etc. In this consideration, the integration of non-stoichiometric semiconductor optical antennas for enhancing 2DLM photodetectors is an appealing research subject.
In 2022, Zhang et al. developed an elaborate design by using a combination of non-stoichiometric nanostructured WO 3-x and 2D Bi 2 O 2 Se nanosheet to realize a high-performance broadband photodetector (Figure 7a). [108] The oxygen-deficient WO 3-x nanostructures are functionalized onto the surface of the Bi 2 O 2 Se nanosheet via an anodic aluminum oxide assisted thermal evap-oration strategy. Importantly, the photocurrent of the hybrid WO 3-x /Bi 2 O 2 Se photodetector is much higher than that of the Bi 2 O 2 Se counterpart device across the vis to NIR spectral range (532-1550 nm) ( Figure 7b). As shown in Figure 7c-e, both responsivity and detectivity of the hybrid WO 3-x /Bi 2 O 2 Se photodetector are markedly boosted as compared to a pristine Bi 2 O 2 Se photodetector upon illumination with wavelengths of 700 and 1310 nm. Specifically, the optimized responsivity and detectivity reach 1.7 × 10 6 A W −1 and 5.4 × 10 11 Jones upon 700 nm illumination. In addition, the optimized responsivity and detectivity reach 48 A W −1 and 7 × 10 7 Jones upon 1310 nm illumination. The improved light-sensing properties are attributed to the plasmonic resonance of the WO 3-x nanostructures. As shown in Figure 7f, upon resonant light excitation, a strong LSPR effect is triggered in the oxygen-deficient WO 3-x nanostructures. Then, the surface plasmons dephase non-radiatively by transferring their energy to free electrons close to the Fermi level, which generates www.advancedsciencenews.com www.advsensorres.com energetic non-equilibrium hot electrons with a maximum energy state (E LSPR ) [181] up to where E F is the Fermi level, and ℏ LSPR is the LSPR energy. Profited from the high kinetic energy, a host of non-equilibrium free carriers can easily overcome the interfacial barrier at the WO 3−x /Bi 2 O 2 Se heterojunction ( B , ≈0.16 eV) to inject into the adjacent Bi 2 Se 2 O nanosheet within an ultrashort time scale of merely a few hundred femtoseconds [182] where D is the density of states of hot electrons, and P is the probability of a hot electron having energy beyond the interfacial barrier. By virtue of the energetic hot electrons and the ultrafast interfacial hot electron injection, the carrier cooling, energy loss, as well as defect capture, are efficiently suppressed. Moreover, the LSPR-induced confinement of an electromagnetic field can prominently enhance the light absorption of the proximal Bi 2 O 2 Se nanosheet. As a consequence, ultrafast light sensing can be realized whilst maintaining high photosensitivity (Figure 7g,h). On the whole, the responsivity and speed of the integrated WO 3−x /Bi 2 O 2 Se photodetector stand out among those of the previously reported Bi 2 O 2 Se-based photodetectors (Figure 7i). [183][184][185][186][187][188][189][190][191] Of note, the free carrier concentration of WO 3−x can be flexibly modulated by the post-synthesis annealing in the hydrogen atmosphere, providing an additional degree of freedom for further engineering the properties of the WO 3−x /Bi 2 O 2 Se photodetector. Thus far, a mass of non-stoichiometric semiconductor nanostructures, such as Cu 3−x P, [98] Fe 1-x S 2 , [21] Cu x S, [101] and MoO x , [106] have been developed as optical antennas for advancing various 2DLM photodetectors. In addition to modifying the surface of 2DLMs via the commonly exploited drop casting and mask/lithography-assisted patterning techniques, non-stoichiometric semiconductor optical antennas can also be integrated through crystal defect-assisted growth of NCs. To this end, in 2019, Hassan et al. prepared substoichiometric Cu 2 S@MoSe 2 heterostructures via Se vacancyassisted growth of Cu 2 S (Figure 8a). [192] As shown in Figure 8bd, the resulting Cu 2 S@MoSe 2 /Si heterojunction photodetector exhibits much-improved photoresponse as compared to a pristine MoSe 2 /Si device. This improvement stems from the synergy of multiple effects. First, the Cu 2 S and MoSe 2 form a pn junction, which is conducive to the spacial separation of photogenerated electron-hole pairs promoted by the built-in electric field. In addition, the defects of MoSe 2 , which are inevitably formed during high-temperature synthesis and are normally scattering and recombination centers for photocarriers, will be passivated by thiol during the above growth process, an additional ingredient contributing to the extended lifetime of photocarriers. Moreover, the copper-deficient Cu 2 S nanostructures exhibit a pronounced LSPR effect, which results in enhanced light-matter interactions. Most recently, by exploiting a similar strategy, Sarkar et al. have reported on a remarkable performance improvement of a MoSe 2 /GaAs vertical heterojunction photodetector by the defect-assisted formation of Cu 2−x S nanostructures on MoSe 2 , [91] consolidating the extensive universality of this strategy. The optimized on/off ratio of the Cu 2 S@MoSe 2 /GaAs heterojunction photodetector reaches 1.1 × 10 5 , which is ≈2 orders of magnitude higher than that of a pristine MoSe 2 /GaAs device (1 × 10 3 ). Moreover, the effective optical window has been markedly broadened from ≈900 to ≈1600 nm with the modification of the Cu 2−x S optical antennas (Figure 8e-g). Herein, the copper-deficient Cu 2−x S NCs play a dual role. On one hand, the intense LSPR effect of the Cu 2−x S NCs results in enhanced NIR light absorption, leading to a substantially broadened effective optical window. On the other hand, Cu 2−x S and MoSe 2 form a pn junction with a favorable type-II band alignment, which promotes high-efficiency spacial separation of photogenerated electron-hole pairs and thus leads to boosted photosensitivity. Of note, this study reveals that the abundant band alignments of semiconductor optical antennas with 2DLMs enable the introduction of a new impetus toward advancing the device properties, which represents a distinctive advantage as compared to the classical metal plasmonic optical antennas.

MXene Optical Antenna Promoted 2DLM Photodetectors
Since the first experimental preparation in 2011, [193] MXenes, a thriving family of transition metal-based 2D materials, have attracted worldwide research attention on account of their intriguing physical properties. Generally, MXenes can be produced by selectively etching away the A layers from the parent MAX phases, where M represents an early transition metal, A represents a group IIIA to VIA element, and X represents carbon/nitrogen. Profited from the high electrical conductivity, MXenes are well known to be promising material platforms for implementing intense LSPR. Thus far, a variety of MXenes, such as Ti 3 C 2 T x , [194] HfTa 4 C 5 , [195] and Nb 2 CT x , [93] have been proven to possess pronounced plasmonic effect, providing flexible choices for optical engineering research. In addition, MXenes serving as plasmonic optical antennas have been widely exploited in multiple domains including solar cells, [196] water splitting, [197] N 2 photofixation, [198] photonic diodes, [199] biosensing, [200] photothermal desalination, [201] and hydrogen evolution reaction. [202] Compared with the previously explored plasmonic optical antennas, MXenes have manifested a wealth of advantages spanning low cost, [203] high air-stability, [204] strong mechanical strength, [205] and easy preparation in large quantities. [206] In addition, unlike heteroatom-doped and non-stoichiometric semiconductors suffering from massive internal crystal defects (i.e., substitution atoms and vacancies), which result in degraded carrier transport properties, enhanced scattering, and expedited recombination of photocarriers, MXenes host relatively intact lattice. Moreover, the resonance spectrum of MXene antennas can be flexibly tailored across a wide spectral range. [207] Most recently, it is predicted that Mo 2 C and Ti 3 C 2 manifest much higher hot carrier generation efficiencies than Au in energy over 1 eV (i.e., the vis to IR band), [208] laying a solid foundation for hot electron devices. Taking advantage of the numerous metrics, MXene optical antennas have also been developed for improving the photosensitivity of various light-sensing devices. [93,209,210] [192] Copyright 2019, American Chemical Society. e) Absorption spectra of pristine MoSe 2 (brown line) and MoSe 2 -Cu 2−x S hybrid (blue line). f,g) Responsivity and detectivity as a function of wavelength of the MoSe 2 /GaAs and MoSe 2 -Cu 2−x S/MoSe 2 /GaAs heterojunction photodetectors. Reproduced with permission. [91] Copyright 2019, American Chemical Society. efficiency (EQE) by more than ten times by embedding Ti 3 C 2 T x nanoflakes into ZnO QD photodetectors. [209] Over the past few years, a series of research advances on exploiting MXene nanostructures as optical antennas for enhancing 2DLM-based light-sensing devices have also been achieved. For example, by using Ti 3 C 2 T x NPs as optical antennas, Zou et al. demonstrated a pronounced improvement in the overall performance of multilayer MoS 2 photodetectors (Figure  9a). [111] In this study, the Ti 3 C 2 T x NPs are prepared by etching out the Al layers from the Ti 3 AlC 2 MAX phase, and they are modified onto MoS 2 via a spin coating strategy. Compared with a pristine MoS 2 device, the responsivity, detectivity, and EQE of the Ti 3 C 2 T x NPs modified MoS 2 photodetector have all been improved across the broad spectral range of 405 to 635 nm (Figure 9b,c). Specifically, the responsivity and EQE have been increased by several times, while the detectivity has been increased by nearly two orders of magnitude. Electromagnetic simulations reveal that the plasmonic effect of the Ti 3 C 2 T x NPs induces a strong elec-tromagnetic field amplification in their vicinity, which results in significant light confinement (Figure 9d-f). Therefore, the optical absorption volume is increased and the light absorption of the Ti 3 C 2 T x NPs/MoS 2 devices has been markedly enhanced, resulting in boosted photosensitivity. Specifically, the optimized responsivity, EQE, and detectivity of the hybrid photodetector reach 20.67 A W −1 , >5000%, and 5.39 × 10 12 Jones, respectively. In addition to being adapted as optical antennas for other 2DLMs, MXenes themselves can also be used as the transport channels of optoelectronic devices. To this end, Velusamy et al. have prepared a high-performance photodetector based on a plasmonic light-sensing channel built of Mo 2 CT x nanosheets via vacuum filtration (Figure 9g,h). [103] Basically, upon illumination with a wavelength matching the LSPR resonance, numerous plasmon-assisted hot carriers are excited. These non-equilibrium carriers are then swept by the source-drain voltage, thus contributing to the generation of photocurrent. Under 660 nm light illumination, the device demonstrates excellent responsivity and  [111] Copyright 2022, American Chemical Society. g) 3D schematic diagram of a Mo 2 CT x photodetector. h) Electron energy loss spectroscopy of Mo 2 CT x . Reproduced with permission. [103] Copyright 2019, John Wiley and Sons. i) 3D schematic diagram of a Ti 3 C 2 T x /GaAs heterojunction photodetector. j) Photoswitching curves upon illuminations with various wavelengths from 405 to 980 nm. k) Energy band diagram illustrating the working mechanism. Reproduced with permission. [107] Copyright 2021, Elsevier Ltd. detectivity of 9 A W −1 and 5 × 10 11 Jones, respectively. On the whole, this study depicts a distinct paradigm for ultra-compact high-performance 2DLM photodetectors.
Beyond the metal-semiconductor-metal type photodetectors, MXene plasmonic optical antennas have also been developed to ameliorate the light-sensing properties of various photodiodes. As an example, in 2021, Zhang et al. achieved high-performance wide-spectrum photoelectric sensing based on a Ti 3 C 2 T x /GaAs Schottky junction photodetector (Figure 9i). [107] In this study, Ti 3 C 2 T x simultaneously serves as an optical antenna and top electrode by leveraging its high electrical conductivity. As shown in Figure 9j, the effective wavelength range of this device covers purple to NIR. It is worth emphasizing that the Ti 3 C 2 T x /GaAs device still exhibits distinct and fast photoswitching behavior under 980 nm light excitation, despite that the photon energy is insufficient to excite the electrons of GaAs (bandgap of ≈1.43 eV, corresponding to a maximum wavelength of ≈867 nm) from the valance band to the conduction band. The substantial broadening of the effective optical window of the Ti 3 C 2 T x /GaAs device as compared to pristine GaAs is mainly ascribed to the plasmonic effect of Ti 3 C 2 T x (Figure 9k). Basically, numerous energetic nonequilibrium hot electrons are generated within Ti 3 C 2 T x through the nonradiative damping of LSPR (i.e., Landau damping). They can easily overcome the Schottky barrier ( SB = M − S , where M is the work function of the metal (i.e., Ti 3 C 2 T x ), and S is the electron affinity of the semiconductor (i.e., GaAs)) at the Ti 3 C 2 T x /GaAs junction to further enter into GaAs [211] where is the hot electron injection efficiency, C F is the Fowler emission coefficient, and ℏ is the photon energy. These nonequilibrium carriers will then be spacially separated by the interfacial built-in electric field, which thus contributes to the photocurrent across the external circuit. Of note, this study exemplifies a distinct avenue to break through the intrinsic limitation on the effective optical window of conventional semiconductors in terms of light sensing, which is imposed by their bandgap values, by leveraging the LSPR effect of plasmonic optical antennas, where the hot carriers are efficiently harvested by the heterointerface before their relaxation. Similarly, in subsequent research work, the same research group has demonstrated ultrabroadband light sensing from purple to MIR (405-3800 nm) by exploiting Ti 3 C 2 T x as the top electrode of a WSe 2 /AlO x /Ge vertical heterojunction photodetector, [109] consolidating the broad universality of the MXene plasmonic optical antennas.
In the context of high metallic conductivity, MXene optical antennas can not only be integrated with 2DLM by surface modification but also can be embedded as electrodes for photoconductivetype 2DLM photodetectors. For example, in 2019, Yang et al. prepared a high-performance InSe photodetector by coupling it with plasmonic Ti 2 CT x grating electrodes (Figure 10a). [105] As shown in Figure 10b-d, the InSe device with patterned Ti 2 CT x (P-Ti 2 CT x ) exhibits higher photocurrent, responsivity, and detectivity than the unpatterned Ti 2 CT x -based device under illumination with different incident light power. Specifically, the optimized responsivity, detectivity, and normalized photocurrent-to-dark current ratio reach 1 × 10 5 A W −1 , 7.3 × 10 12 Jones, 3.5 × 10 13 W −1 , making it stand out in the rank of the state-of-the-art 2DLM photodetectors (Figure 10e). The elevated properties are associated with the plasmonic resonance-induced strong light-matter coupling from the Ti 2 CT x nanogratings, which results in enhanced light absorption (Figure 10f).
Of note, compared with the MXene nanostructures with random morphology, thickness, and structure, the elaborately patterned MXene antennas can enable the precise tailoring of the plasmonic resonance absorption range by modulating the geometric parameters of the nano-/micro-structures, which holds an indisputable prospect for broadband light sensing. To this end, Jeon et al. have systematically explored the MoS 2 photodetectors coupled to the Mo 2 C grating electrodes with various periodicities (Figure 10g). [104] In this study, the Mo 2 C is produced via a carbonization approach, where MoS 2 is treated in a hybrid CH 4 /H 2 /Ar atmosphere under high temperatures. The Mo 2 C gratings are patterned by electron beam lithography and plasma etching. Benefiting from the intense plasmonic resonance, the MoS 2 photodetector coupled with periodic Mo 2 C stripes exhibits increased optical absorption, consequently, the responsivity and light-to-dark current ratio (LtDR) are both significantly higher than those of a pristine MoS 2 device (Figure 10h). As shown in Figure 10i, the photocurrent of the hybrid MoS 2 /Mo 2 C photodetector is significantly higher than that of the pristine MoS 2 device across the spectral range of 405-1310 nm. Furthermore, the spectral photocurrent of the hybrid MoS 2 /Mo 2 C photodetector with different grating periods exhibits photocurrent peaks in different wavelengths, indicating that the spectral selectivity is closely related to the geometry of the Mo 2 C grating, providing an additional degree of freedom for tailoring the photoresponse of 2DLM photodetectors. It is to be emphasized that the MoS 2 /Mo 2 C photodetector even manifests pronounced photoresponse in the spectral range beyond 800 nm, where the photon energy is insufficient to excite electrons of MoS 2 from the valance band to the conduction band. This is a benefit from the synergy of dual factors, the low Schottky barrier (≈70 meV) at the MoS 2 /Mo 2 C heterointerface, and the strong plasmonic effect of the Mo 2 C grating. In principle, the energetic hot carriers generated by the non-radiative decay of the surface plasmons of metallic Mo 2 C can easily surmount the MoS 2 /Mo 2 C interfacial barrier and then inject into MoS 2 , thus contributing to the photocurrent and triggering the sub-bandgap photosensitivity beyond the intrinsic long-wave limit ( Max ) of MoS 2 (i.e., Max (nm) = 1240 |E g | , where E g is the bandgap value in electron volt). From this, by integrating Mo 2 C strips with multiple grating periods, high photosensitivity over a broad spectral range (405-1310 nm) has been realized, which is ascribed to the well-complementary optical properties of antennas with various sizes. Of note, in a subsequent study, Yin et al. reported on MXene-contact enhanced broadband light sensing in GeS photodetectors, [212] demonstrating outstanding general applicability.

Summary
As a concluding remark, non-noble plasmonic optical antennas can prominently enhance the light-matter interactions and promote electron-hole pair spacial separation in their vicinity through LSPR by coupling and trapping freely propagating plane waves into an adjacent semiconductor as the typical noble metal optical antennas, whereas their production cost is generally much more competitive. In addition, the resonance spectrum of non-noble plasmonic optical antennas can be widely tuned and it is far beyond the scope of classical noble metal antennas, enabling the substantial broadening of the effective optical window and the breakthrough of development bottleneck of classical noble metals. Thus far, a number of non-noble plasmonic optical antennas, which are categorized as non-noble metals, heteroatom doped semiconductors, non-stoichiometric semiconductors, and MXenes in this review, have been developed to tailor the lightsensing performance of 2DLM photodetectors, and remarkable Figure 10. a) 3D schematic diagram of InSe photodetectors with unpatterned (1-2) and patterned (2-3) Ti 2 CT x electrodes. b-d) Photocurrent, responsivity, and detectivity as a function of incident power of InSe photodetectors with unpatterned (hollow circles) and patterned (solid circles) Ti 2 CT x electrodes under 405 (blue circles), 655 (red circles), and 785 (brown circles) nm illuminations. e) Comparison of the InSe/P-Ti 2 CT x photodetector with state-of-the-art 2DLM based photodetectors. f) Absorption spectra of unpatterned Ti 2 CT x (black line) and patterned Ti 2 CT x (red line). Reproduced with permission. [105] Copyright 2019, American Chemical Society. g) 3D schematic diagram of a MoS 2 photodetector with a Mo 2 C grating electrode. h) Responsivity (solid) and LtDR (hollow) as a function of wavelength of pristine MoS 2 (squares) and hybrid MoS 2 /Mo 2 C (circles) photodetectors. i) Photocurrent as a function of wavelength of MoS 2 photodetectors without (blue line) and with (green and red lines) Mo 2 C grating electrodes. Reproduced with permission. [104] Copyright 2019, John Wiley and Sons.
achievements have been made ( Table 1). In this regard, these light-confining materials hold the indisputable potential to take the place of classical noble metal plasmonic optical antennas toward implementing the next generation of optoelectronic devices, which will bring about further performance breakthroughs of 2DLM photodetectors.

Outlook
Despite significant progress, the research on non-noble plasmonic optical antenna-promoted 2DLM photodetectors is still in its infancy, and many fundamental issues need further exploration.

Developing Novel Non-Noble Plasmonic Antennas
In addition to the above-mentioned optical antennas that have been extensively exploited for 2DLM photodetectors, a number of neoteric non-noble plasmonic active competitors with intense LSPR effects have also emerged in the past few years including carbon nanotube, [213] graphene, [214,215] NbSe 2 , [64] TaSe 2 , [216] WTe 2 , [217] GeSe, [218] TiS 2 , [219] borophene, [220] NiSe, [92] Adv. Sensor Res. 2023, 2, 2200079 CoSe, [92] MoN, [221] TiN, [222,223] ZrN, [224] Sb 2 Te 3 , [112] Bi 2 Te 3 , [225] silicon, [226,227] germanium, [228] bismuth, [229] tellurium, [230] GaP, [231]  NPs. [112] In another study, Ge et al. revealed highly sensitive detection of bilirubin in serum and urine samples by capitalizing ultrathin TaSe 2 as a surface-enhanced Raman scattering layer. [232] Compared with the previously reported optical antennas, these newly emerged non-noble plasmonic candidates have exhibited unique advantages. As proof, NbSe 2 is a layered van der Waals material with a dangling-bond-free surface, [64] which enables its epitaxy-free on-chip integration with various 2D photosensitive building blocks, where the heterogeneous interface is naturally strain-free and defect-free. This is conducive to mitigating the undesirable interfacial recombination of photocarriers. Furthermore, with the advantage of the ultrathin nature, the plasmonic resonance property of the few-layer NbSe 2 can be flexibly regulated across a substantially broad spectral range of ≈360 cm −1 through the mature electrostatic gating (Figure 11a-c). This theoretically provides an additional degree of freedom for the post-fabrication active tuning of the photoresponse of 2DLM photodetectors coupled to the atomically thin NbSe 2 antennas. In another study, it is demonstrated that the plasmon resonance of ultrathin TaSe 2 strips can cover an ultra-broadband spectral range from the terahertz region (40 m) to the telecom region www.advancedsciencenews.com www.advsensorres.com (1.55 m), which can be further modulated by tailoring their thickness and dielectric environments. [216] Such spectral scope is far beyond those accessible by conventional noble metal plasmonic nanostructures, enabling it with excellent compatibility with the longwave light-sensing 2DLMs such as MnBi 2 Te 4 , [233] black phosphorus, [234] PtSe 2 , [235] PtSe 2 , [173] PdSe 2 , [236] PtTe 2 , [237] Bi 2 O 2 Se, [45] ZrGeSe, [238] etc. Beyond this, 2DLMs have grown as a huge material library with a substantially broad distribution of bandgap sizes (0-6 eV). [4] Profited from the naturally passivated surface enabling circumvention of the strict lattice-matching constrains for covalent semiconductors, 2DLMs with complementary bandgap values and non-noble optical antennas with complementary LSPR spectra can be arbitrarily assembled within a single device, which enables a potential scheme for high-performance full-spectrum light sensing system toward superintegrated circuits.

Optimization of Fabrication of Non-Noble Plasmonic Antennas
Thus far, a substantial portion of non-noble plasmonic optical antennas has been prepared by the problematic solution methods. [95,102,111] Although these synthesis technologies are cost-efficient and easily scalable, reaction precursors, ligands, and dispersants are commonly used, which will inevitably bring residues. These residues will seriously deteriorate the carrier transport properties, resulting in a certain lag in the practical device performance as compared to the theoretical ceiling. For example, one critical dilemma of the colloid plasmonic optical antennas is the long insulating alkyl chain ligands on the surface, which will substantially block the charge transfer from antennas to 2DLMs, as well as give rise to charge trapping. To address this issue, Gong et al. have proposed a ligand exchange protocol where the insulating long-chain octadecylamine (ODA) ligands are replaced by the conducting short-chain 3-mercaptopropionic acid (MPA) ones (Figure 11d-f). [99] On the other hand, apart from the surface residuals, the repeatability of the synthesis of nonnoble plasmonic optical antennas remains an additional predicament to be overcome. Therefore, the immediate next step is to develop contamination-free and economical dry synthesis technologies.

Fine Control of Resonant Properties of Non-Noble Plasmonic Antennas
Currently, many of the studies on non-precious plasmonic optical antennas suffer from the drawback of high randomness. [102,108] That is, the researchers usually explore what they garner from the synthesis. As is well known, the optical properties of plasmonic nanostructures are closely related to their component, [239] morphology/shape, [240,241] and size. [242,243] For example, in 2016, Kim et al. revealed that the crystalline anisotropy played a critical role in the plasmonic property of Cs-doped WO 3 , which uniquely caused strong LSPR band-splitting into two distinct peaks with comparable intensities. [244] In 2019, Liu et al. developed cation-exchange reactions to selectively introduce various aliovalent atoms into metal-oxide nanocrystals, realizing programmable LSPR via component engineering (Figure 11g). [245] In another study, Li et al. demonstrated that the LSPR band of W 18 O 49 underwent a gradual blue shift and an intensity increase as their morphology changed from nanowires to nanobundles and to urchin-like nanospheres. [177] These pioneering advances have important implications for the development of hybrid nonnoble antenna/2DLM photodetectors. However, there still lack sophisticated and controllable means for precisely engineering the parameters of non-noble optical antennas. In this respect, more efforts should be devoted to controllably customizing the constituent, shape, and size of non-precious plasmonic optical antennas in the upcoming future. As an attempt, Kapetanovi et al. reported on the fabrication of a triangle ensemble of ITO microstructure by using electron beam lithography and radiofrequency sputtering. [246] In another study, Yu et al. demonstrated the construction of an Al-doped ZnO microdisk array by lithography-assisted thermal evaporation. [110] More progress in this field can be expected.

Polarization Sensing Enabled by Low-Symmetry Non-Noble Plasmonic Antennas
Limited by the intrinsic high crystal symmetry, photodetectors based on many 2DLMs, such as the widely explored group 6 transition metal dichalcogenides, can only detect the light intensity, whereas the polarization state is unable to be identified. [247] In a commercial scheme, a photodetector and a polarizer are conjointly used, which, however, suffers from large device volume. To achieve compact integration, it is thus necessary to develop 2DLM photodetectors intrinsically sensitive to the polarization state of incident light. Coupling low-symmetry optical antennas is a potential strategy to address this issue. As proof, by decorating elliptical Au nanodisks onto a few-layer MoS 2 photodetector, Chen et al. have realized polarized light sensing with a dichroic ratio of 1.45 (Figure 11h-j). [248] The emergence of anisotropy is mainly ascribed to the low-symmetry elliptical Au nanodisks, which result in distinct near-field distribution under light irradiation with different polarization directions. As a consequence, polarization-dependent light absorption is triggered, which thus results in polarization-resolved photocurrent. In principle, the idea of integrating low-symmetry antennas can be flexibly applied to non-noble plasmonic optical antennas, and further exploration is essential in the upcoming future. Figure 11. a) Schematic illustration of the configuration for tunable NbSe 2 plasmon. b) Plasmonic profiles under various gate voltages. c) Resonant wavenumber as a function of gate voltage. Reproduced with permission. [64] Copyright 2021, John Wiley and Sons. d) Schematic illustration of the ligand exchange avenue. e,f) The calculated electron localization function in molecular dynamics simulation in the cases of FeS 2 with long ODA ligand and short MPA ligand, respectively. Reproduced with permission. [99] Copyright 2021, John Wiley and Sons. g) Schematic illustration of the ion exchange avenue. Reproduced with permission. [245] Copyright 2019, Nature Publishing Group. h) 3D schematic diagram of a MoS 2 photodetector coupled with elliptical Au nanodisk arrays. i) Electric field distribution in the vicinity of an elliptical Au nanodisk under 0°(top) and 90°(bottom) polarized light irradiation. j) Polar plots of the photocurrent as a function of polarization angle of pristine MoS 2 (blue line), Au nanocircles modified MoS 2 (green line), and Au nanoellipses modified MoS 2 (red line) photodetectors. Reproduced with permission. [248] Copyright 2020, John Wiley and Sons.
In summary, the integration of non-precious plasmonic optical antennas has provided a distinct pathway for the further breakthrough of the device performance of 2DLM photodetectors, endowing them with the potency to take the place of devices built of covalently bonded bulk light-sensing materials. Gratifying advances in this realm have been realized, but there also exist some critical challenges in the current stage. Provided these issues can be reasonably resolved, it is believed that 2DLMs will play a pivotal role in the next-generation optoelectronic industry.