InAs / GaSb Type-II Superlattice Detectors

InAs/(In,Ga)Sb type-II strained layer superlattices (T2SLs) have made significant progress since they were first proposed as an infrared (IR) sensing material more than three decades ago. Numerous theoretically predicted advantages that T2SL offers over present-day detection technologies, heterojunction engineering capabilities, and technological preferences make T2SL technology promising candidate for the realization of high performance IR imagers. Despite concentrated efforts of many research groups, the T2SLs have not revealed full potential yet. This paper attempts to provide a comprehensive review of the current status of T2SL detectors and discusses origins of T2SL device performance degradation, in particular, surface and bulk dark-current components. Various approaches of dark current reduction with their pros and cons are presented.

While MCT detectors have very large quantum efficiency (>90%) and detectivity, they are still plagued by nonuniform growth defects and a very expensive CdZnTe substrate that is only available in limited quantities by a foreign manufacturer.There has been significant progress on development of MCT on silicon substrates, but good performance has been limited to the MWIR band only.Moreover, MCT is characterized by low electron effective mass resulting in excessive leakage current [7].The InSb detectors do not cover the LWIR spectral range.QWIPs are based on III-V semiconductors and their mature manufacturing process enables them to be scaled to large format FPAs with a high degree of spatial uniformity [8][9][10].However, due to polarization selection rules for electron-photon interactions in GaAs/AlGaAs QW, this material system is insensitive to surface-normal incident IR radiation resulting in poor conversion quantum efficiency, In addition, their large dark currents lower the operating temperature and increase the operating cost of the imager.The development of FPAs based on mature III-V growth and fabrication technology and operating at higher temperatures will result in highly sensitive, more reliable, lighter, and less costly IR sensors than currently available ones.
The InAs/(In,Ga)Sb T2SL material system is characterized by a broken-gap type-II alignment schematically illustrated in Figure 1 with electron and hole wavefunctions having maxima in InAs and GaSb layers, respectively.The overlap of electron (hole) wave functions between adjacent InAs (GaSb) layers result in the formation of an electron (hole) minibands in the conduction (valence) band.Optical transition between the highest hole (heavy-hole) and the lowest conduction minibands is employed for the detection of incoming IR radiation.The operating wavelength of the T2SLs can be tailored from 3 m to 32 m by varying thickness of one or two T2SL constituent layers [11][12][13].Some parameters of T2SL constituent materials, InAs and GaSb, are shown in Table 1.[21,22] 0.220 [23,24] InGaSb layers of InAs/InGaSb T2SL are subjected to biaxial compression strain causing splitting of light hole and heavy-hole minibands in the T2SL band structure and, therefore, suppression of Auger recombination rates relative to bulk MCT detectors [25,26].However, the majority of the research in the past ten years has focused on the binary InAs/GaSb system.This is attributed to the critical thickness limitations imposed on strained material grown with the large mole fraction of In.The scope of this paper is also limited to the InAs/GaSb T2SL devices.

Characterization of T2SL Material
System.The physics behind the T2SL material system is not yet very well understood.Different theoretical methods have been applied to understand the band structure, electronic, and optical properties of superlattices.For example, Flatté et al. have undertaken extensive theoretical modeling of the band structure of superlattices [27][28][29], including investigation of electronic structure of dopants [30].Features of T2SL photoabsorption spectra and optical properties of T2SL detectors were studied by Livneh et al. [31] and Qiao et al. [32], respectively, using k ⋅ p tight-binding model [33].Empirical pseudopotential method, in its canonical shape [34][35][36] and four-component variation that includes interface layers [37], was successfully utilized for the heterojunction design of T2SL devices.Bandara et al. [38] have modeled the effect of doping on the Shockley-Read-Hall (SRH) lifetime and the dark current; Pellegrino and DeWames [39] have performed extensive modeling to extract the SRH lifetime from darkcurrent measurements.
Background carrier concentration is one of the fundamental properties of the absorber layer of T2SL detector since it determines the minority carrier lifetime and diffusion lengths.Transport measurements in T2SL are difficult because of the lack of semi-insulating GaSb substrates.Several techniques have been reported to measure and analyze the electrical properties of T2SL by different groups.Magneto-transport analysis [40] was performed on T2SL structures grown on top of electrically insulating AlGaAsSb buffer in order to suppress parasitic conduction.Hall [41], capacitance-voltage, and current-voltage measurements [42] of T2SL structures grown on semi-insulating GaAs substrate directly or with the interfacial misfit (IMF) dislocation arrays technique [43] were also reported.Variable magnetic field geometric magnetoresistance measurements and a mobility spectrum analysis, (MSA) technique for data analysis, have been employed by Umana-Membreno et al. [44] to study vertical minority carrier electron transport parameters in T2SL structures.Works of Christol et al. [45,46], Haugan et al. [47], and Szmulowicz et al. [48,49]are concerned with the influence of T2SL composition and growth conditions on background carrier concentration and mobility.
Since performance of T2SL device is strongly dependent on T2SL structural perfection, the information on interfacial roughness, compositional profile (i.e., interfacial intermixing), and interfacial bonding across the noncommon anion layers of InAs/GaSb T2SL is very important.Growth conditions of T2SLs have been optimized by various research groups to improve the interface quality [50][51][52][53][54]. Steinshnider and colleagues [55][56][57][58] utilized the cross-sectional scanning tunneling microscopy (XSTM) to identify the interfacial bonding and to facilitate direct measurements of the compositional grading at the GaSb/InAs heterojunction.In situ study of origins of interfacial disorder and crosscontamination in T2SL structures [59,60] revealed importance of Arsenic (As) background pressure control during the GaSb layers growth.Luna et al. [61] proposed the method of systematic characterization of InAs-on-GaSb and GaSb-on-InAs interfaces in T2SL with resolution less than 0.5 nm.

Limitations of T2SL Technology
Despite the numerous technological and theoretically predicted advantages T2SLs offer over present-day detection technologies, the promise of superior performance of T2SL detectors has not been yet realized.The T2SL detectors are approaching the empirical benchmark of MCT's performance level, Rule 07 [93]; however, the dark-current density demonstrated by the T2SL detectors is still significantly higher than that of bulk MCT detectors, especially in the MWIR range, as illustrated in Figure 2.
To understand the reasons of high dark-current levels demonstrated by the T2SL detectors the origins of dark current have to be analyzed.Generally, dark current in detectors based on narrow band gap semiconductors may be differentiated into "bulk" and "surface" currents.The most important "bulk" dark currents are (i) generation-recombination (G-R) current associated with the SRH process in the depletion region of the detector and (ii) thermally generated diffusion current associated with Rogalski [94] or radiative process in both the n-and p-extrinsic regions of the detector [95].The SRH G-R process occurs through the trap levels within the energy gap thus limiting lifetime of the minority carriers.The origins of SRH centers are not well understood.According to the statistical theory of the SRH process, the SRH rate approaches a maximum as the energy level of the trap center approaches midgap.Thus, the most effective SRH centers are those located near the middle of the band gap [96].Analysis of the defect formation energy of native defects dependent on the location of the Fermi level stabilization energy has been performed by Walukiewicz [97], who reported that, in bulk GaAs and GaSb, the stabilized Fermi level is located near either the valence band or the midgap, whereas in bulk InAs the stabilized Fermi level is located above the conduction band edge.From this observation, the midgap trap levels in GaAs and GaSb are available for SRH recombination, whereas in InAs they are inactive for the SRH process, suggesting a longer carrier lifetime in bulk InAs than in bulk GaSb and GaAs materials.Experimentally measured values of carrier lifetimes yielded ∼325 ns for bulk InAs and ∼100 ns for bulk GaSb, thereby confirming the initial observation [98].It may then be hypothesized that native defects associated with GaSb are responsible for the SRHlimited minority-carrier lifetimes observed in InAs/GaSb T2SL.
Several methods have been employed to measure lifetime of photogenerated carriers of T2SL, including optical modulation response [38,47], time-resolved PL [99][100][101][102], and photoconductive response variation measurements [103], to name just a few.Some of them provide more direct measures of lifetime, while others rely on assumptions or further measurements to perform extraction of lifetime [39,104,105].Overall, the lifetimes reported for MWIR and LWIR InAs/GaSb T2SL range from 0.13 ns [106] to about 100 ns [107].These values are significantly lower compared to the MCT devices operating in the same wavelength range [107].
A "surface" dark current component is associated with the surface states in the junction.During the individual pixel isolation process, the periodic crystal structure terminates abruptly resulting in formation of unsatisfied (dangling) chemical bonds at the semiconductor-air interface responsible for generation of surface states within the band gap and pinning of the Fermi level.Moreover, etch by-products, surface contaminant associated with the fabrication procedure, and differential etching also create additional interfacial states contributing to the dark current.Scaling of the lateral dimensions of a T2SL detector (e.g., typical mesa dimensions of individual FPA pixels are 20 m × 20 m) makes FPA performance strongly dependent on surface effects due to a large pixel surface/volume ratio.
This paper aims to review various ways of improving performance of T2SL detectors in order for T2SLs to be the technology of choice for high-performance IR imaging systems.Proposed solutions for the reduction of "bulk" and "surface" dark-current components as well as improvement of detector signal-to-noise ratio and operating temperature limits will be discussed in detail.

Proposed Solutions for the Improvement of T2SL Detector Performance
3.1.Reduction of "Bulk" Dark Currents.To overcome the carrier lifetime limitations imposed by the GaSb layer in an InAs/GaSb T2SL, the type-II Ga-free SL, that is, InAs/InAsSb SL, may be utilized for IR detection.A significantly longer minority carrier lifetime has been obtained in an InAs/InAsSb SL system as compared to an InAs/GaSb T2SL operating in the same wavelength range (at 77 K, ∼412 ns, and ∼100 ns, resp.)[100,108].Such increases in minority carrier lifetimes, along with demonstrated band gap adjustability [109] and suppressed Auger recombination rates [110], suggest lower dark currents for InAs/InAsSb SL detectors in comparison with their InAs/GaSb T2SL counterparts.However, performance, in particular, signalto-noise ratio, of InAs/InAsSb SL-based detectors with pin [111] and nBn [112] architectures was not superior to T2SLbased devices operating in the same wavelength range.This may be attributed to the increased tunneling probability in InAs/InAsSb SL system due to the smaller band offsets [111] and significant concentration of SRH centers in this material [113].
Thermally generated diffusion currents may be significantly suppressed by the incorporation of barriers into conduction and valence bands to impede the flow of carriers associated with dark current (noise) without blocking photocurrent (signal).The improved performance of these T2SL devices is credited to better confinement of the electron wavefunctions, reduced tunneling probability, increased electron effective mass in modified T2SL structures, and reduction in dark-current through the use of current blocking layers that reduce one or more dark-current component.nBn [86,87], pBiBn [12], M-structure [83], W-structure [84], CBIRD [70], and N-structure [85] are examples of T2SL detectors with barrier architecture.
The band-offset tunability is critical parameter for the realization of barrier devices.Barrier layers are selected such that the hole-blocking layer offers an unimpeded electrons flow while blocking holes and electron-blocking layer fulfill the opposite function.Hence, one requires hole-(electron) blocking layers to have zero valence (conduction) band offsets with the absorber layer.Moreover, for the efficient barrier structure design, complete macroscopic simulations are required to get a good assessment of actual dark current and photocurrent.This simulation may as well help with design optimization of barrier structures, in particular, selecting an optimal barriers thickness, composition, and doping concentration.
The extension of concept of heterostructure barrier engineering in T2SL resulted in realization of interband cascade infrared photodetectors (ICIPs) [114][115][116][117][118]. In ICIP detectors each cascade stage is comprised of an absorber region, relaxation region, and interband tunneling region.While photocurrent is limited to the value produced in an individual absorber, adding of extra-stages benefits the signal-to-noise ratio, since the noise current in such devices scales inversely with the total number of stages.Ability to change number of stages with different absorber thicknesses is important for the design of T2SL detectors with maximized signal-tonoise ratio.The drawbacks of ICIPs are associated with the complicated structure of these devices.In particular, due to the number of layers and interfaces in the structure, some of the fundamental device physics is still unclear and MBE growth procedure is challenging.
Ability to heteroengineer the band structure of the T2SL devices stipulates realization of one more type of low-noise T2SL detectors, avalanche photodiodes (APDs) [119][120][121].Control of individual layer thickness and composition offers great flexibility in engineering of the electron band structure to initiate single-carrier ionization.Moreover, APDs with either electron or hole dominated avalanching may be fabricated by engineering the higher lying T2SL energy levels.It should be noted that an APD device with hole dominating avalanching is expected to have lower noise due to reduced tunneling of heavier holes.Existence of hole dominated avalanching structure also opens up possibility of combining separate electron and hole multiplication regions in a single device achieving very high gain with low excess noise factor.

Reduction of "
Surface" Dark Currents.Despite numerous efforts of various research groups devoted to the development of effective passivation schemes for T2SL detectors, there is still no well-established and generally acknowledged procedure for passivation of such devices.Part of the problem is the complexity of T2SL system, composed by the hundreds of relatively thick (several monolayers, MLs) InAs, GaSb, and, sometimes, AlSb layers, and thin (typically, less than 1 ML) interfacial GaAs and InSb layers [55,57,58,122].Passivation should satisfy dangling bonds of all these T2SL constituent materials, originated at exposed device sidewalls after mesa definition process, and prevent formation of interface states in the T2SL band gap.
The great advantage of T2SL system, band gap tunability, allowing realization of detectors spanning wide IR range, serves as a disservice for the passivation development.Interface states cause the pinning of Fermi level with the bands bend towards lower energy near the surface.This band bending induces accumulation or type inversion of charge resulting in surface tunneling currents along sidewalls.As was shown by Delaunay et al. [123], the narrow band gap devices (LWIR and VLWIR, with band gap of 120 meV or lower) are more susceptible to the formation of charge conduction channels along the sidewalls.Consequently, the same passivation may be suitable for the T2SL MWIR and inefficient for the T2SL detectors with longer operating wavelength.
Moreover, passivation should exhibit thermal and long term stability.In other words, passivation layer must not undergo any change in its constitutional, physical, and interfacial properties at variable temperatures (30-300 K) during the lifetime of the T2SL detector (typically, 10,000 hrs).Finally, passivation has to be easily integrated into the FPA fabrication process.
In addition, since passivation applied on rough surfaces, or surfaces contaminated by native oxides, and foreign particles will result in little or no improvement of device performance, we spent some time discussing the surface preparation issues.To achieve minimal surface leakage, the device sidewalls must be smooth, with no patterns of preferential etch presented, and clean, with removed native oxides and etch by-products.Moreover, vertical etch profile is essential for the realization of high-fill factor, small pixel pitch, and large format T2SL FPAs.The thorough comparisons of various surface preparation and passivation techniques of T2SL detectors are out of the scope of this review article and can be found in literature [124,125].Next two sections aim to familiarize the reader with various mesa definition and passivation methods developed for T2SL devices.

Surface Preparation.
Definition of nearly vertical mesa sidewalls that are free of native oxide and defects is the crucial step in InAs/GaSb T2SL detector fabrication process [126,127].Presence of elemental antimony on the etched T2SL device sidewalls [128] may result in the conduction channel parallel to the interface, which leads to increasing of surface component of dark current.Unwanted native oxides are usually removed prior to or during the pixel isolation process with immersion in ammonium sulfide [127], phosphoric or hydrochloric acid based solutions [129].Introduction of BCl 3 gas into the plasma chemistry is also effective in removal of native oxides and redeposited by-products [130].
Nowadays, high-density plasma etch processes are commonly utilized for InAs/GaSb T2SL material in spite of inevitable degradation of sidewall surface electronic properties due to ion bombardment or unwanted deposition of etch by-products [131,132].Plasma chemistry usually consists of chlorine-based precursors (BCl 3 , Cl 2 , or SiCl 2 ) due to high volatilities of gallium, indium, antimony, and arsenide chlorides providing fast etch rates and smooth morphologies [133].The resulting etch profiles are vertical due to the plasma sheath and the ionized gas directionality.Damage produced during the dry etch may be partially restored by subsequent chemical treatment [134].Due to the ability of wet etches to cause virtually no surface electronic damage, a chemical etch attracts attention of researchers for single-pixel T2SL device fabrication [135][136][137][138][139][140].However, the isotropic nature of wet etch process resulting in concave sidewall profile and an unavoidable tendency to undercut etch masks making precise dimensional control more difficult stipulates limited application of wet etches for T2SL FPA fabrication.

Passivation.
Conventional passivation methods of T2SL devices include encapsulation of device sidewalls, by thick layer of dielectric or organic material, and sulfidization.Dielectric passivation of T2SL detectors is compatible with current T2SL FPA fabrication procedures and, consequently, very appealing to the T2SL scientists and engineers.Numerous reports on passivation of MWIR and LWIR T2SL detectors by silicon oxide or silicon nitride have been published over the last fifteen years [74,81,130,[141][142][143][144].Dielectric passivation, though shown to be effective, presents the challenges of developing high-quality, low fixed, and interfacial charges density dielectrics at process temperatures substantially lower that the InAs/GaSb T2SL growth temperature to prevent the T2SL period intermixing.Moreover, native fixed charges presented in dielectric passivation layer can either improve or deteriorate the device performance [143]; consequently, the dielectric passivation may not passivate the low band gap materials as effectively as high band gap materials.
The simplest sulfidization scheme of T2SL detectors is device immersion in aqueous solution of ammonium sulfide.No native oxide removal step is required prior to passivation because the native oxides are etched by (NH 4 )OH formed in water solution of ammonium sulfide.Short-term benefits for the MWIR and LWIR T2SL device performance have been reported [127,129,163] and the necessity for a suitable capping layer to preserve good passivation quality in the long term was reaffirmed.Thioacetamide (TAM, C 2 H 5 NS) [124,164] and octadecanethiol (ODT, CH 3 [CH 2 ] 17 SH) [85] treatments offer formation of more stable bonds between sulfur and T2SL constituent elements (Ga, In, As, and Sb) compared to weaker III (V)-oxygen-S bonds formed after ammonium sulfide treatment.
One of relatively new sulfidization methods is electrochemical passivation (ECP) [129,165] that is saturation of dangling bonds with sulfur through electrolysis in Scontaining solution.Though effective, sulfur layer deposited through ECP may oxidize easily and additional encapsulation is required.Electron-beam evaporated ZnS satisfies the dangling bonds with S-atoms simultaneously acting as an encapsulant [119][120][121]166].
Recently, several research groups reported the "combined" approach for the passivation of T2SL detectors.For example, Zhang et al. [167] noticed that the anodic sulfide passivation combined with the SiO 2 significantly improved the performance of MWIR T2SL detectors.DeCuir Jr. et al. [147] found that the sulfide chemical treatment followed by the SU-8 treatment inhibits the formation of native surface oxides, satisfies dangling bonds, and prevents the sulfide layer degradation over time.

Other Methods of T2SL Detector Performance Improvement
The bulk components of the dark current (SRH and thermally generated diffusion current) in T2SL detector may be significantly diminished by scaling thickness of the device.The abridged quantum efficiency (QE) of such device may be restored through plasmon assisted coupling of incident electromagnetic radiation while maintaining low dark-current level.Transmission enhancement and QE increase through subwavelength metal hole array [168] and corrugated metal surface structure [169], respectively, have been reported for MWIR T2SL detectors.Surface currents may be suppressed by reduced exposure of narrow gap materials to the environment, for example, as a result of encapsulation of etched sidewalls with wide band gap material [133,170] or buried architecture [84] that isolates the neighboring devices but terminates within a wider band gap layer.The former passivation approach requires very careful surface cleaning prior the overgrowth procedure, whereas latter is subjected to the possible crosstalk issues in FPAs due to the uncertainty of the lateral diffusion length of minority carriers.If the values of lateral diffusion length are larger than the distance between neighboring pixels in the FPA, crosstalk between the FPA elements can be encountered that leads to the degradation of image resolution.
Another approach for the realization of high performance T2SL sensors is growth of T2SL structures on high-index plane GaSb [171].The thickness of the T2SL detector grown on the GaSb (111) substrate is reduced due to the natural difference of lattice parameters in the ( 111) and (100) directions, whereas heavy hole confinement is increased by a factor of three [172].This translates into thinner detector structures for a given detection wavelength and absorption coefficient realized on (111) GaSb substrate, resulting in shorter growth times.This also means decreased costs and material usage, both of which are highly desirable.Moreover, the decreased detector volume results in an improved signal-to-noise ratio, since the number of thermally generated carriers is correspondingly reduced.

Summary
This work provides a review of the current status and limitations of IR detectors based on an InAs/GaSb T2SLs.It should be noted that applications of T2SL system are not limited to the IR detection only.Low thermal conductivity of T2SL identifies it as a prospective material for low-temperature Peltier coolers [173].Spatially separated confinement of electrons and holes, signature of type-II band alignment, initiated InAs/GaSb core-shell nanowires realization [174,175].Field-effect transistors (FETs) [176,177] and thermo photovoltaic (TPV) [178] T2SL devices are another examples of unconventional applications of T2SL material system.
Despite the numerous theoretically predicted advantages that T2SLs offer over MCT, InSb, and QWIP-based detectors, intensive heterostructure engineering efforts and development of epitaxial growth and fabrication techniques, the promise of superior performance of T2SL detectors has not been yet realized.The dark-current density demonstrated by the T2SL detectors is still significantly higher than that of bulk MCT detectors, especially in the MWIR range.
The complexity of T2SL system, along with the intricate detector architectures, results in no universal solution for the suppression of dark currents.Different approaches that address suppression of either bulk or surface dark current components in order for T2SL to be the technology of choice for high-performance imaging systems have been presented.
The SRH and thermally generated diffusion currents may be significantly reduced by exclusion of GaSb layer from InAs/GaSb T2SL stack, that is, Ga-free T2SL, and by the incorporation of barriers device structure to impede the flow of carriers associated with dark current (noise) without blocking photocurrent (signal), respectively.Passivation treatment of the exposed device sidewalls decreases the surface currents.However, development of effective passivation technique is hindered by the ease of native oxide formation and requirements to the etched surface.In addition, the same passivation may be suitable for the T2SL MWIR and inefficient for the T2SL detectors with longer operating wavelength.Finally, one of the most effective passivation approaches, saturation of unsatisfied chemical bonds with sulfur atoms, results in formation of passivation layer with poor long-term stability, and additional encapsulation is required.
Integration of T2SL detectors with surface plasmon couplers and utilization of high-index plane GaSb substrates are recent alternatives for the improvement of T2SL detector performance.Despite the promising preliminary results, both of these directions require additional investigation.
In conclusion, unique combination of band structure engineering flexibility and material properties of InAs/GaSb T2SL provide a prospective benefit in the realization of next generation IR imagers.Performance of MWIR and LWIR T2SL detectors has not achieved its theoretically predicted limit.To fully realize the T2SL potential methods of suppression of various dark current components have to be developed.Up-to-date techniques of dark current reduction include not only traditional passivation, but advanced heterostructure engineering and integration of T2SL with nanostructures as well.

Table 1 :
Some band structure parameters for InAs and GaSb (0 K).