Emerging technologies for high performance infrared detectors

3.1.1 Intersubband QWIP

Figure 2 shows the low-temperature energy bandgap diagram of different semiconductors with diamond and zinc-blende structure versus their lattice constant matching. Thin layers of semiconductors with close lattice constants – within each of the gray zones – can be grown on top of each other, using strain compensation. Such thin QWs and barriers can be designed as shown in Figure 3 to produce localized quantum states within each conduction or valence band, with a specific energy difference to resonate with the targeted IR radiation. Such “intersubband” photodetectors are called QWIPs.

Figure 2:

The low-temperature energy bandgap diagram of different semiconductors with diamond and zinc-blende structure versus their lattice constants by Nobel Laureate Esaki [41].

Figure 3:

Schematic of the conduction energy band profile for a GaAs/AlGaAs QWIP under zero (above) and finite (below) bias.

Among the different types of QWIPs, GaAs/AlGaAs QW detectors are the most mature due to the almost perfect natural lattice match between GaAs and AlGaAs [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54]. GaAs/AlGaAs QW devices have many advantages, including their ability to be used with standard GaAs-based manufacturing techniques and processing technologies, highly uniform and well-controlled molecular beam epitaxy growth on six-inch GaAs wafers, high yield and thus low cost, more thermal stability, and extrinsic radiation hardness. However, the optical cross-section absorption is limited by the dopant concentrations, leading to a low quantum efficiency and relatively poor performance at temperatures >40 K due to short intersubband lifetimes (dominated by fast phonon relaxation). While some complex quantum structures have been explored [51], the emerging ideas to address this performance limitation are based on optical antenna and plasmonic structures, which will be discussed in Section 4.

3.1.2 Interband superlattice IRPD

A rapidly maturing III–V IRPD structure is InAs/In_xGa_1−x Sb (InAs/InGaSb) strained-layer T2SL detectors [52], [53], [54], [55]. The optical absorption of T2SL occurs in interband transition, which makes it an intrinsic device with high quantum efficiency. T2SLs can produce a bandgap that is smaller than constituting semiconductors, due to the staggered (type II) energy band lineup of these layers. However, this band lineup causes low overlap integral between the electron and hole wavefunctions, and hence a low IR absorption. In recent years, T2SL has been further improved by a current blocking layer which has larger energy bandgap than the surrounding SL layers [56], [57]. An example of this new band diagram can be seen in Figure 4.

Figure 4:

Schematic diagram of W- and M-T2SL LWIR photodiodes.

(A) Band profiles of W-T2SL [58] and (B) p-doped M-T2SL structure [59].

Most of the blocking barrier consists of aluminum incorporated into either the InAs or GaSb layer to form the band structure period of the SL either like a W or an M shape. However, the precise position of the barrier in the middle of the well is needed to increase the number of energy states available at the desired energy level [57], [59]. Razeghi’s group [58], [60], [61], [62] showed a blocking barrier T2SL with detectivity as high as 1.05×10¹² cm Hz^1/2/W at 150 K in the MWIR regime and then a 320×256 pixel FPA was constructed using this design, with imaging possible up to 170 K.

3.2.1 QDIP based on III–V material

In 1998, the first QDIP based on intersub-level transitions was demonstrated along with the first QD laser [70]. QDIPs are similar to QWIPs but with extra advantages due to the three-dimensional (3D) confinement in the QDs compared to the 2D confinement in QW. There are three major advantages of QDIPs compared to QWIPs: (1) QDIPs are intrinsically sensitive to IR irradiation at normal incidence due to the breaking of the polarization selection rule [71], (2) QDIPs have lower dark current than QWIPs because of weaker thermionic emission from the QDs with 3D quantum confinement of carriers [72], and (3) the discrete energy levels in QDs have no dispersion, which reduces the phonon scattering and can lead to longer carrier lifetime (>100 ps), and so higher operating temperatures [73], [74]. The detection mechanism in QDIPs is also based on intersubband transitions between the quantized energy levels of the dots and continuum states [63], [75]. However, QDIPs have lower absorption quantum efficiency due to the small fill factor area of QDs and large inhomogeneous broadening of the self-assembled QDs [63], [75]. This inhomogeneous broadening also makes fine tuning to specific wavelengths difficult, and leads to issues with strain and dislocations with increasing number of absorption layers, diminishing the absorption coefficient.

In the last few decades, Lin et al. [76] improved the performance of QDIP responsivity, detectivity, as well as demonstrating higher operating temperatures and lower dark currents by using sandwiched QD layers with Al_0.3Ga_0.7As current blocking barriers. The background-limited infrared photodetector (BLIP) of the improved QDIP was demonstrated to be as high as 250 K [77]. Lin et al. [78] showed that the QDIP can be further improved in regard to the responsivity and lower dark current by direct doping the QDs. Later on QDIPs with QW active regions, known as DWELL structures, were proposed to improve the detection wavelength tunability with InGaAs strain relief layers [79], [80]. Surprisingly, DWELL IRPDs turned out to have advantages such as enhancement of absorption, low dark current, and high responsivity [81], [82], [83]. DWELL IRPDs have been extensively explored by Krishna’s research group, and some of the main achievements are presented in their review articles [84]. They have demonstrated some impressive results including: (1) multiwavelength operation in the MWIR, LWIR, and VLWIR, (2) fine tuning operating wavelength, (3) tunable bias, and (4) higher operating temperatures, especially for the VLWIR response. A 320×256 LWIR FPA was developed and images were obtained at 78 K [84]; however, this FPA has low responsivity and quantum efficiency.

3.2.2 QDIP based on Ge/Si heterostructures

Germanium QDs can be grown on Ge/Si and Ge/SiGe/Si multilayer heterostructures using the effect of the self-assembly of semiconductor nanostructures during the heteroepitaxial growth of materials with a large lattice mismatch [85]. Similar to other materials, Ge QDIP improves the performance of the heterostructures in multiple ways, such as by allowing absorption of normally incident electromagnetic radiation, high (up to 1000) photoelectric gain, and lower dark currents in Ge/Si QDIP [86]. High temperature operation QDIP at mid-IR (MIR) has also been achieved, even up to room temperature [87]. However, the quantum efficiency and photocurrent of Ge/Si QDIP are far lower than those of the bulk material. Despite the low quantum efficiency, Ge/Si QDIP is still one of the most popular QDIPs because of its lower cost. This is due to well-established silicon technology for fabricating MIR devices with low defect content and a high spatial homogeneity of photovoltaic characteristics. Ge/Si QDIPs also have better thermal expansion coefficients, which can easily be matched to a silicon readout circuit so that the array size is not limited by the cooling-induced mechanical stresses of the entire structure.

3.3.1 Photoconductor NW IRPD

NW photoconductors are NW-based photodetectors where the material is sandwiched between two electrodes [97]. NW photodetectors based on III–V compound semiconductors are one of the most studied due to their excellent transport properties, easy fabrication, and wide tunability wavelength by alloy bandgap engineering. Also, their low capacitance leads to high operating speeds. For example, an intersecting array of InP NWs can have a fast photoconductive response (14 ps FWHM at 780 nm) [98]. In recent years, InGaAs and InAs have been studied intensely. InGaAs has potential bandgap tunability from the NIR to MIR, as well as high electron mobility and small leakage current. Single-crystal In_0.65Ga_0.35As nanowires are reported to have a large responsivity of 6.5×10³ A/W over a broad range from 1100 to 2000 nm [99]. The photoresponsivity of 5.3×10³ A/W was reported in Schottky-Ohmic contacted InAs NW photodetectors as shown in Figure 6 at the wavelength around 1.5 μm [100]. II–VI material NWs, such as CdS, CdSe, ZnSe, CdTe and ZnTe, are typically used for visible light detection. CdS NWs and nanobelts are among the most studied photoconductors in group II–VI for visible and ultraviolet light detection [89]. In the visible-NIR range (400–800 nm), CdTe nanoribbons with p-type conductivity present significant photoresponse to irradiation with a high responsivity of 7.8×10² A/W [101].

Figure 6:

Schottky Ohmic contacted InAs NW photodetectors.

(A) Schematic of InAs Photoconductor NW operating at visible to NIR regime. (B) The IV characteristics under different illumination wavelengths at the same power density (0.75 mW/mm²) [100].

3.3.2 Phototransistor NW IRPD

A phototransistor is a bipolar or unipolar transistor in which light is converted to carriers and the current can be amplified through transistor action, leading to much greater photosensitivity. However, most of the NW phototransistors today have surface recombination loss, as carriers recombine before they can be collected. This problem can be solved by a core/shell design. A good example is an InAs NW phototransistor with a self-assembly photogating layer containing randomly distributed lattice defects as a trapping layer to capture photoexcited electrons [102]. This core/shell-like n-InAs NW phototransistor has a very large photoconductive gain of 10⁵ and a fast response time of 12 ms at the wavelength of 1.2 μm [102]. Single GaAsSb NWs have also shown a decent responsivity of 2.37 A/W with a low operating bias voltage of 0.15 V achieved at the wavelength of 1.3 μm [103]. These GaAsSb NWs were grown in a self-assembly fashion using a horizontal flow metalorganic vapour phase epitaxy reactor at 100 mbar.

3.3.3 Heterostructure NW IRPD

A lot of promising heterostructured NW architectures as shown in Figure 7 adopted methods from thin-film compound semiconductor photodetector technologies such as homogeneous, inhomogeneous, Schottky junction photodiodes, and avalanche photodetector. The p-n/p-i-n junction is normally used for homo/heterojunction NW photodiodes which operate under reverse bias. A single NW of InAs/InAsP axial heterojunctions is reported to have very low dark current, strong polarization dependence, and high photoconductivity when operated at 77 K [108]. Another example of NW heterostructures is core-shell NW junctions which attract a lot of interest due to high efficient photosensing and high light absorption enhancement [109]. For IR detection, GaAs systems are the most widely investigated. Examples include GaAs/AlGaAs core-shell [104], [110], [111], [112], [113] and GaAs/InGaP/GaAs core-multishell NWs [114]. Among them, an NIR photodetector based on a core-shell GaAs/AlGaAs NW exhibited a peak photoresponsivity of 0.57 A/W at room temperature, comparable to large-area planar commercial GaAs photodetectors, and a high detectivity of 7.2×10¹⁰ cm Hz^1/2 W⁻¹ at λ=855 nm [115].

Figure 7:

Different types of semiconductor NW: (A) GaAs/AlGaAs (core-shell) [104], (B) MSM based on CdS nanowire [105], (C) high-resolution TEM image of an InP nanowire with an embedded InAsP quantum dot [106] and (D) GaAs-based avalanche photodiode (APD) nanoneedles [107].

Another example of heterostructure NWs is Schottky photodiode structures [usually the construction is an metal semiconductor metal (MSM) photodetector], which exhibit fast response speed [105]. Arguably, the most promising NW photodetectors are the ones based on avalanche multiplication. NW avalanche photodiodes (APDs) with diverse configurations have been demonstrated. These include the “crossed” n-CdS/p-Si NW heterojunction [116], axial p-i-n single Si NW homojunction [116], InAsP QD in the axial InP p-n junction [117], radial GaAs nanoneedle junction [107], and radial GaAs p-n junction [118]. GaAs-based APD GaAs nanoneedles exhibit an excellent multiplication factor as large as 29 at −2 V and 263 at −8 V [118].

3.3.4 Si/Ge NW IRPD

Si/Ge NW IRPDs naturally have indirect bandgap which limits it from the high optical absorption application. However, by reducing their dimension, one can increase the quantum confinement and change the band structure. Also, impurity doping can lead to better performance photodetectors. A detailed review of Si/Ge NWs and their applications has been published recently by Ray et al. [119]. However, most of the works on Si/Ge NWs are focused at visible wavelength and very limited applications on NIR and MIR. The most notable work on Si NW in the NIR region with a high responsivity of 2.5×10⁴ A/W in NIR wavelength (900 nm) at a zero bias condition has been reported recently [120]. The Si NWs with 80–100 nm diameter were fabricated by metal (Ag)-assisted electroless chemical etching techniques. Further enhancement of the Au NP-decorated Si NW has also been reported by Jee et al. [121]. The Au NP-decorated Si NWs achieved 63 times enhancement at the wavelength of 1000 nm, but the reported responsivity is around 0.1 A/W.

3.4.1 Graphene-based photodetector

Graphene is a promising material for photodetectors compared to conventional semiconductors due to ultrahigh mobility, making it suitable for high-speed communications [127], [128], [129], [130]. Single-atomic-layer graphene is stable, low cost, easy to fabricate, and has high internal quantum efficiency [131] when used in the NIR and MIR regions for optoelectronic applications. However, single-layer graphene is an inherently weak light absorber, which originates from its short interaction length [132], [133], [134], and the narrow effective area of lateral graphene p-n junctions also limits the efficiency of photocarrier extraction [135], [136], [137]. In other words, the applications of grapheme-based photodetectors are limited by the lower external quantum efficiency and photoresponsivity in comparison to traditional photodetectors [130].

Various approaches have been proposed to enhance the sensitivity, including the introduction of a bandgap or an electron trap layer, such as building graphene QD-like structures [138] or cutting graphene into nanoribbons [139]. However, these bandgap opening methods degraded the electronic performances of graphene, especially the high mobility, thus leading to a decrease in the photoresponse speed [138]. Unfortunately, the overall quantum efficiency of graphene is quite low compared to conventional materials such as III–V semiconductors, due to the nature of zero bandgap structures. Figure 8A summarized the responsivity versus the response time for detectors based on difference 2D material, and comparison with commercial silicon and InGaAs photodiodes. Figure 8B also showed the wavelength dependency of different 2D material use in photodetector.

Figure 8:

Summarization of important characteristics for detection application using different 2D materials.

(A) Responsivity versus the response time for detectors based on 2D material, and comparison with commercial silicon and InGaAs photodiodes (black squares). (B) Bandgap of the different layered semiconductors and the covered electromagnetic spectrum. The exact bandgap value would depend on the number of layers, strain level and chemical doping. The asterisk indicates that the material’s fundamental bandgap is indirect [152].

3.4.2 TMDC-based photodetector

Recently, another family of 2D materials with far better optical properties has emerged. Single-layer TMDCs [145], such as MoS₂ and WSe₂, have been intensely researched. The direct bandgap of monolayer TMDCs leads to efficient light emission covering the energy range from below 1 eV to well above 2.5 eV [153] and beyond, making them promising for a wide range of optoelectronic devices. Bandgap tunability with the layer thickness is another important property of this material. The bandgaps of trilayer, bilayer, and monolayer MoS₂, determined by photoluminescence, change from 1.35 to 1.65 and 1.8 eV [154], respectively. The bandgap of other typical TMDCs, such as WSe₂, MoSe₂ [146], WS₂ [147], and GaTe [148], also increases with decreasing layer thickness, due to the quantum confinement of carriers in the direction normal to the 2D plane. 2D TMDC photodetectors exhibit decent responsivity from the IR to the near UV range. The first reported monolayer MoS₂ photodetector [149] exhibited a responsivity of 7.5 mA/W in the visible range. Choi et al. [150] demonstrated a responsivity of ~100 mA/W by using a multilayer MoS₂ photodetector. However, the speed of photoresponse is relatively slow (ranging from microseconds to seconds) owing to the trapping of photocarriers. It is also noticeable that most of the high-sensitivity TMDC-based photodetectors reported above are for visible and NIR application. However, the recently re-discovered few-layer black phosphorus (bP) is an interesting material for photodetection for IR and far infrared (FIR), especially due to its intermediate bandgap between graphene (i.e. zero bandgap) and TMDCs. The study by Youngblood et al. [151] shows an intrinsic responsivity of up to 135 mA/W and 657 mA/W in 11.5-nm- and 100-nm-thick devices, respectively, at room temperature and at a frequency of 3 GHz. The few-layer bP is embedded within an on-chip waveguide structure with a few-layer graphene top gate, allowing for optimal interaction with light and tunability of the carrier density. The bP-based device presented in [151] stands out from the rest by showing comparable performance to both a commercial silicon photodiode and a graphene-based detector. Despite the large amount of funding and research invested on 2D materials, there is currently very limited set of 2D material covering the IR region of the spectrum [148]. This might represent an important future research area, considering the strong need for conformal and thin IR detectors.

3.4.3 2D material-based heterostructure IRPD

As discussed in the previous section, conventional graphene and TMDC-based phototransistors have relatively poor responsivity (~10⁻² A/W) owing to their weak optical absorption. Recently this limitation has been addressed by 2D material synthesized with other semiconductor material, nanostructured or different 2D materials. For example, MoS₂ nanocrystal on silicon on insulator structure [155], [156], MoS₂/bp heterojunctions [157], graphene/MoTe₂/graphene heterostructures [158], graphene/MoS₂ heterojunctions, and WSe₂/MoS₂ heterojunction photodetectors [157] show significant improvement in responsivity to the visible and NIR regions. A recent review paper on different dimensional material heterostructures [159] summarizes the works on different dimensional material for increasing the responsibility of the 2D material in the IRPD. One of the most notable works is QD on the graphene photodetector, where the QD optical absorption leads to a large photoexcited charge carrier in the graphene channel, and added with the high mobility of graphene, the QD/graphene heterojunction photodetector achieves a massive responsivity exceeding 10⁷ A/W in the NIR region [160]. However, most of the mixed-dimensional heterojunction photodetectors discussed above have a relatively slow response (10¹–10⁻⁶ s) due to their enhanced capacitance and charge traps. Nonetheless, there are a growing number of active research studies on faster IRPD devices with lower responsivity based on 2D heterojunctions. For example, single walled carbon nanotubes/MoS₂ p-n heterojunction photodetectors show responsivity in the visible and NIR with a response time of 15 μs [161]. Similarly, field-effect transistors based on the heterojunction between a monolayer of WS₂ and PbS CQDs demonstrated a photoresponsivity of ~14 A/W, while exhibiting a fast photoresponse time of ~153 μs. Most of such 2D heterojunction-based devices have the potential to become low-cost and high-performance IRPDs, while significant future research is needed to the commercialization stage.

3.4.4 CQD

In the past decade, CQD as shown in Figure 9 has emerged as one of the most appealing potential nanomaterials for IRPDs. CQDs show quantum confinement and are commonly made of II–VI, III–V, and IV–VI semiconductors by means of inexpensive and scalable wet-chemical synthetic procedures [163], [164]. Photodetection by CQD is based on the conductivity of CQD solid films. The first generation of CQD light detectors focused on photoconduction [165], where photogeneration of an electron-hole pair is followed by capture of one type of carrier (can be the electrons or holes), while the other remains substantially free to travel. Most of the CQD photoconductors reported until now are for the NIR to MIR regime, and based on the lead chalcogenide nanocrystals [166], [167]. The speed of these CQD photoconductor devices depends on the carrier lifetime. Very large photoconductive gain, in the range of 10³–10⁶, can be achieved [166]. Photodetectors based on CQDs exploit tuning of the band edge to suitably chosen spectral cutoff energy, enabling dark current minimization for a given desired spectral response. The dark current is amplified via the same gain mechanism as photocurrent. However, it could be minimized by using photo-field-effect transistors [168]. The transistor introduces a threshold, so that turn-on occurs only after a certain photocarrier density (hence light level) is exceeded. PbSe CQD field effect phototransistors based on metallic (Au/Ag) nanowire transparent electrodes with a high responsivity of 2×10⁴ A/W and a high specific detectivity of 7×10¹² Jones at low operation voltage (~1 V) in the NIR have been reported [169].

Figure 9:

An image from the PbS CQD-based focal plane array (FPA) camera operating in the SWIR band.

The CQD-based camera can see through a silicon wafer. The first inset shows the image taken using a normal visible camera. Second inset image is the PbS CQD-based focal plane array (FPA) structure [162].

The speed of CQD detectors has been historically an issue. The extended lifetimes of trapped photocarriers, which are the source of slow response, can be controlled via the CQD surface chemistry and surface passivation [167]. Extracting photocarriers by using the built-in field in a photodiode provides a ready path to achieving efficient photocurrent generation; when the excited-state lifetime is tipically ~100 ns, a built-in or applied potential on the order of 1 V can achieve a drift-based extraction time comparable with the excited lifetime when the lower of the electron or hole mobility exceeds 2×10⁻³ cm²/V s in the MIR regime [170]. Similarly, in the MIR regime, the CQD photodiode based on monodispersed HgSe CQD photodetectors that are illuminated with radiation resonant with intraband electronic absorption between 3 and 5 μm is reported. The HgSe CQD photodetector achieved the lowest dark current of 80 μA, but with a relatively small detectivity of 8.5×10⁸ Jones at 80 K [171].

The maturity of the CQD in imaging has been much faster compared to other nanostructures. PbS CQD-based FPAs and camera prototype devices have been demonstrated, with a wide range of array and pixel sizes. The imaging arrays display a spectral response from 400 to 2000 nm, a high dynamic range of ~60 dB, a fast rise/fall time of about 2 μs, and a very respectable detectivity ~10¹² Jones [162], [172]. These FPAs display multispectral sensitivity with detectivity comparable to InGaAs devices that currently dominate the commercial market. However, they offer a very attractive low-cost route, since they can be spin-coated on top of silicon readout integrated circuit (ROIC) chips. This approach could in principle produce a high-yield and low-cost alternative to the current indium bump-bonding of the sensor chip to the ROIC chip.

4.1.1 SWS for AR

Most of the semiconductor materials used for IRPDs have a large refractive index (typically 3–4). Therefore, AR structures are important for IRPDs, since the loss due to reflection from the surface is around 25–40%, and even higher at increased incident angles. The simplest AR coating can be made by depositing one or several optically thin films on the surface utilizing interference effects to cancel the back-reflection [173]. However, this approach has significant limitations due to the change of reflectivity with the incident angle. This is a rather fundamental limitation in thin-film interferometric filters, since the wavenumber (k-vector) parallel to the films is preserved throughout all layers. Another related issue is the narrow spectral response band of this approach, which renders such AR coating layers ineffective in modern “multi-color” and “multi-band” IRPDs. Antireflective condition can be generated by forming SWSs in front of the detector [174]. Such structures are also denoted as high spatial frequency structures or subwavelength surface textures [175]. SWSs generally have a two-fold function in light management in photodetectors: since they effectively represent gradient-index structures, they produce a broadband impedance matching that leads to a broadband AR performance. SWSs can also be designed to function as efficient mode couplers that couple the incident mode to many modes inside the high-index detector volume, and with a diverse polarization. This property can be used to couple to material with specific polarization, such as QWIP. The optical behavior of SWSs can be satisfactorily approximated in a majority of situations by using the effective medium theory [176].

SWS structures can be formed randomly across the surface or as periodic arrays. Random SWS structures give broadband antireflective properties, but repeatability is an issue. On the other hand, periodic SWSs can be designed based on required wavelengths, but usually the fabrication is more complex and costly. Random SWSs can be formed by self-assembled nanospheres to perform nanosphere lithography by reactive-ion etching and obtain patterns of nanocavities on interface material [177]. For example, a cone-shaped Si nanostructure can reduce the reflectance (a planar SI) from 30% to 3% by using inductively coupled plasma – reactive ion etching at the NIR wavelength (800–1000 nm) [178]. This irregular structure is also referred to as random surface corrugations. Similarly for the periodic SWS, periodic structure lithography (holographic lithography, e-beam lithography) is performed followed by chemical or physical etching. SWSs offer important advantages over interference AR coatings. Examples of period SWSs include moth eye-like structures, 2D pyramidal, and also nanopillars. Figure 10 shows that SWSs can effectively enable fabrication of any desired effective refractive index profile [179]. These SWS nanostructures can efficiently reduce the reflection of the surface leading to higher absorption in a wide range of wavelengths including the NIR and MIR regions [180].

Figure 10:

Predicted transmission of MWIR light propagating from air through an AR texture into silicon for various profile structures [179].

4.1.2 Resonant cavity structure

One the most commonly used light-trapping structures in IRPDs is the resonant cavity structure (Figure 11). The absorption within the detector can be significantly increased by placing the active area into a resonant cavity between resonator mirrors (usually consisting of a highly reflective mirror and a distributed Bragg mirror). If the product of the effective absorption coefficient and the active area thickness is small enough, the cavity will maintain a single allowed optical mode, while the other modes will be suppressed [182]. The resonant cavity enhanced (RCE) photodetector is usually known as RCEPD.

Figure 11:

3D schematic illustration, top view microscope image and cross-sectional scanning electron microscope image of the fabricated RCEPDs [181].

The amplification of the radiation at the resonant wavelength in an RCE device and rejection of others results in an increased spectral selectivity of resonant detectors. However, the RC structure also decreases the speed of the detector [183] which becomes limited by carrier transit time. Another disadvantage is an increase of the allowed operating temperature because of the shift of the BLIP limit [184]. RCE structures were used to improve performance in almost all IRPD systems such as p-i-n diodes [185], QW [181], T2SL [186], QD [187], quantum dot in quantum well [188], and light emitting diodes [189].

4.1.3 PC structure

Research on PC structures with a periodic refractive index modulation has opened up several ways to control light. Most existing devices are realized as 2D PC structures, as they are compatible with standard semiconductor processing. PC behaves as a high reflection coefficient mirror for all wavelengths within the photonic bandgap (PBG), and the waves with such wavelengths are evanescent within it. An important property of PC is defect modes, i.e. the allowed energy modes within their PBG which appear when the symmetry of the PC is broken [190], [191].

The use of a PC cavity to improve characteristics of IRPDs was described in [192]. The photodetector is enclosed within a PBG structure with a PBG either equivalent to the electronic bandgap of the utilized semiconductor material or larger than it. A defect may be introduced in the PC on the incident side of the photodetector, and its transmission peak adjusted to a desired wavelength. Such an approach improves the performance of the detector in two ways at the same time. As other methods of photodetector enhancement through optical path increase, it increases the probability of absorption of useful radiation in the active region and thus improves quantum efficiency and the responsivity. At the same time, reabsorption (photon recycling) increases the radiative lifetime and thus shifts the BLIP detectivity limit. The photoresponse of the PCS-QWIP (Figure 12) shows a wider response peak but additionally displays several pronounced resonance peaks [193].

Figure 12:

PCS-QWIP design.

(A) SEM image of a cleaved PCS and (B) cross-section through the PCS-QWIP structure [193].

4.2.1 Plasmonic enhancement in IRPD

Surface plasmon-mediated light-trapping schemes for photodetectors [199], [200] may be divided into two distinct groups; namely, (1) localized surface plasmon consists of plasmonic nanoparticles or nanovoids, and (2) surface plasmon polariton consists of different diffractive structures [197]. The former includes field localization and generation of hotspots [195], plasmon-based singular optics [199], and metamaterial-based transformation [196]. The latter includes structures such as gratings [198] for field coupling into guided modes and subwavelength plasmonic crystals that may be periodic [201] or quasi periodic [194].

Among subwavelength PCs described within the context of IR photodetection, an important position belongs to 2D arrays of nanoapertures in opaque metal film. Such structures first drew attention for their ability to transmit light in spite of the dimensions of nanoapertures being much smaller than the operating wavelength. This effect is commonly known as the extraordinary optical transmission (EOT) [198]. EOT is a result of the excitation of surface plasmon polaritons at the metal surface. It forces the electromagnetic waves incident to the surface to pass through the apertures. The transmission couples all incident electromagnetic energy on the surface to the mode impinging through the aperture. Thus, large field localizations appear in the apertures. This behavior effectively corresponds to impedance matching between propagating waves and the perforated metal film.

Interesting structures of these types similar to Figure 14 have been developed, to enhance the weak optical absorption of graphene. Compared to traditional structures including cavities [205], [206] and waveguides [202], [207], these metal plasmonic structures [203], [204] can enhance the light-matter interaction over a broad spectral band. A planar microcavity based on graphene sandwiched between two highly reflecting mirrors achieved a 26-fold enhancement in the absorption of graphene with a responsivity of 21 mA/W [205]. These are also used for other 2D materials [202], [207]. The third approach is to enhance the local electrical field, which is guided into the region of the graphene photoactive area using plasmonic metal structures, hence increasing the light absorption [203], [204]. This field enhancement is proven to dramatically improve the performance of graphene-based photodetectors. The graphene photodetectors can couple with metal nanostructures at different resonance frequencies, and the photocurrent and external quantum efficiency could be enhanced by 1500% [204].

Figure 14:

Photodetection enhancement by metallic plasmonics and graphene.

(A) Schematic of a monolayer graphene sheet is on top of a silicon bus waveguide [202]. (B) Different metallic structure to enhance the performance of grapheme-based device. Scale bar=1 μm [203]. (C) Schematic diagram of a grapheme-based photodetector with asymmetric metal contacts (Ti and Pd) decorated with Au nanostructures. (D) Photocurrent profile with laser scanning from outside of the Ti electrode (black line: normal graphene device; red line: plasmon resonance-enhanced graphene device). The inset shows the band profile of the asymmetric metal-contacted graphene photodetector [204].

4.2.2 Optical antenna

Nanoantenna or optical antenna [194] is a structure that couples the propagating light waves and the evanescent (near) fields. The large change of the photon momentum is the unique property of such devices. Depending on the nanoantenna type and its design, the mode confinement in the evanescent field can be deeply subwavelength. The simplest and the most basic nanoantenna is a nanosphere. This structure actually behaves like a dipole and its scattering properties can be calculated using the Mie theory [208]. Much larger field localizations are obtained using two identical nanospheres (known as dimer) or in an array (nanoparticle chain) [209]. Many other shapes can be used as nanoantennas such as nanorods, two-wire [210] bowtie [211] diablo-type nanoantenna [212], the Yagi-Uda, group of director nanorods [213], spiral nanoantennas [214], and those with fractal geometries [215]. Many other shapes can be used.

In 2010, our group reported an ultra-thin QWIP integrated with a nanohole-array optical antenna (Figure 15) [217]. The optical antenna captures far-field light and enhances its concentration at the active region of QWIP. The polarization selection rule in intersubband QWs prevents absorption of normal-incident light. The nanohole-array antenna can efficiently convert the polarization of normal-incident light to a perpendicular polarization by exciting the surface plasmons on the surface of the perforated metal. The peaks of responsivity and detectivity both occur at 8 μm, reaching 7 A/W and 7.4×10¹⁰ cm Hz^1/2/W, respectively. Such a high responsivity and detectivity is due to a high quantum efficiency (QE) of 40%, which is an order of magnitude larger than the QE of a plain QWIP structure with a similar thickness of ~λe QE.

Figure 15:

An ultra-thin QWIP integrated with a nanohole-array optical antenna.

(A and B) SEM images of nanohole array optical antenna integrated on a QWIP photodetector. (C) The measured responsivity of the device at bias 0.7 V at 78 K [216].

Several similar structures were reported since then, including a plasmonic optical antenna coupled with a LWIR QDIP [218]. The antenna consists of a pair of half-wavelength dipole antennas with a 20 nm gap in between. E-field enhancement of more than 30 times is obtained in the gap, corresponding to over 900 times light intensity enhancement. This LWIR QDIP shows strong incident-angle-dependent photocurrent enhancement with a clear antenna-induced detection pattern (i.e. directivity).

In the last decade, high-gain and high-directivity IRPDs have been intensely studied. Optical antennas coupled with metal-oxide-metal (MOM) diodes have been used to demonstrate beam steering in the MIR (10 μm) [219]. The devices show narrow beam widths of ~35° FWHM in power, and reception angles of ±50°. Two dipole antennas coupled to coplanar strip waveguides, inspired by similar systems in microwave electronics [220], are coupled to a rectifying MOM diode. Due to the interference between the two antennas, the diode can be placed off-center and “see” off-normal angles. Recently, a double-metal nanoantenna as microcavity has been fabricated on a QWIP detector to confine light in 3D [221]. Strong subwavelength light confinement of this microcavity antenna leads to three times reduction in dark current due to the reduced detection area. The structure achieved a peak responsivity of 0.074 A/W photoconductive gain g=0.25 at a wavelength of 9 μm at all incident angles.

Recently, our group proposed a novel metallo-dielectric hybrid antenna, with a metallic cavity antenna coupled to a photonic jet produced by a dielectric microsphere. This structure as shown in Figure 16 has a high QE of about 50% and a high directivity of about 20 dB. The quantum efficiency and the directivity are significantly better than the bowtie optical antenna with ~1% QE and poor directivity gain (a few dBs) [222]. When integrated with QWIP, the detectivity of the hybrid antennas is more than 30 times higher than the limit of conventional QWIP detectors at all temperatures due to the strong directivity gain. However, the optical antennas cannot increase the detectivity of the interband detector beyond the limit of conventional devices, because of their comparable radiative and non-radiative lifetime. Another similar optical antenna integrates a split-ring bull’s eye antenna on a germanium detector to focus light from a relatively large area into a subwavelength germanium slab. The measured photocurrent is improved by seven times, with a responsivity of 3.93 mA/W at 1310 nm [223].

Figure 16:

Schematic of the hybrid antenna and FDTD simulation cross-sectional view of the metallo-dielectric micro cavity.

The right figure shows the optical power flow (arrows) and normalized power consumption density (optical absorption density) throughout the cavity at the operating wavelength λ~8 μm [222].

4.2.3 Metamaterial structures

Subwavelength dielectric and metallic structures made in 1D, 2D, and 3D arrays have created a novel class of structures known as electromagnetic metamaterials [224], [225]. They could be defined as artificial structures whose effective electromagnetic properties cover regions beyond what is found in nature. It is possible to obtain effective relative dielectric permittivity, relative magnetic permeability, and refractive index in a vastly expanded range by tailoring their geometry and material properties (example as shown in Figure 17).

Figure 17:

Scanning electron microscopy images of the integrated IRPD with a metamaterial structure.

The photocurrent spectra at different polarization angles are presented [216].

For example, negative index metamaterials are combined with dielectrics to produce a broad range of new device properties [216], [226], [227], [228], [229]. Metamaterial-containing multilayers exhibit many peculiar properties, with both theoretical and practical interests. For example, it is possible to fabricate resonant cavities with subwavelength dimensions [230]. Such structures might be used to reduce the volume of the absorber, and hence the dark current/noise, while maintaining a high QE. Another application that is related to IRPDs is a multilayer metamaterial based on the so-called double fishnet structure with enhanced IR absorption and a relatively large directivity [231]. Compared with the aforementioned photonic jet approach, this method is all planar, and uses standard micro/nano-fabrication methods. A persisting problem with such metamaterials is their large absorption losses – specifically in their negative refractive index range, as well as relatively narrow bandwidths due to their resonant nature.