Heterostructure infrared photovoltaic detectors

Infrared Physics & Technology 41 (2000) 213±238

www.elsevier.com/locate/infrared

Heterostructure infrared photovoltaic detectors Antoni Rogalski * Institute of Applied Physics, Military University of Technology, 2 Kaliskiego Street, 00-908 Warsaw 49, Poland Received 18 February 2000

Abstract HgCdTe remains the most important material for infrared (IR) photodetectors despite numerous attempts to replace it with alternative materials such as closely related mercury alloys (HgZnTe, HgMnTe), Schottky barriers on silicon, SiGe heterojunctions, GaAs/AlGaAs multiple quantum wells, InAs/GaInSb strained layer superlattices, high temperature superconductors and especially two types of thermal detectors: pyroelectric detectors and silicon bolometers. It is interesting, however, that none of these competitors can compete in terms of fundamental properties. In addition, HgCdTe exhibits nearly constant lattice parameter which is of extreme importance for new devices based on complex heterostructures. The development of sophisticated controllable vapour phase epitaxial growth methods, such as MBE and MOCVD, has allowed fabrication of almost ideally designed heterojunction photodiodes. In this paper, examples of novel devices based on heterostructures operating in the long wavelength, middle wavelength and short wavelength spectral ranges are presented. Recently, more interest has been focused on p±n junction heterostructures. As infrared technology continues to advance, there is a growing demand for multispectral detectors for advanced IR systems with better target discrimination and identi®cation. HgCdTe heterojunction detectors o€er wavelength ¯exibility from medium wavelength to very long wavelength and multicolour capability in these regions. Recent progress in two-colour HgCdTe detectors is also reviewed. Ó 2000 Elsevier Science B.V. All rights reserved.

1. Introduction Recent success in applying infrared (IR) technology to remote sensing problems has been made possible by the successful development of highperformance IR detectors over the last ®ve decades. Many materials have been investigated in the IR ®eld. Interest has centred mainly on the wavelengths of the two atmospheric windows, 3±5 and 8±14 lm, though in recent years, there has been increasing interest in longer wavelengths stimulated by space applications.

*

Fax: +48-22-685-9109. E-mail address: [email protected] (A. Rogalski).

Recently, more interest has been focused on p±n junctions heterostructures [1,2]. Photodiodes with their very low power dissipation, easy multiplexing on focal plane silicon chip and less stringent noise requirements for the readout devices and circuits, can be assembled in two-dimensional (2-D) arrays containing a very large (106 ) number of elements, limited only by existing technologies. Systems based upon such focal plane arrays (FPAs) can be smaller, lighter with lower power consumption, and can result in much higher performance than systems based on ®rst generation detectors [3,4]. Photodiodes can also have less low-frequency noise, faster response time, and the potential for a more uniform spatial response across each element. However, the more complex processes

1350-4495/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 1 3 5 0 - 4 4 9 5 ( 0 0 ) 0 0 0 4 2 - 6

214

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

needed for photovoltaic detectors have an in¯uence on slower development and industrialisation of the second generation systems, particularly for large arrays. 2. Fundamental limitation to photodiode performance The photodetector is a slab of homogeneous semiconductor with the actual ``electrical'' area Ae that is coupled to a beam of infrared radiation by its optical area Ao (Fig. 1). Usually, the optical and electrical areas of the device are the same or close. The use of optical concentrators can increase the Ao =Ae ratio. The current responsivity of the photodetector is determined by the quantum eciency g and by the photoelectric gain g. The idea of photoconductive gain g was put forth by Rose [5] as a simplifying concept for the understanding of photoconductive phenomena and is now widely used in the ®eld. The photoelectric gain is the number of carriers passing contacts per one generated pair. This value shows how well the generated electron±hole pairs are used to generate the current response of a photodetector. Both values are assumed here as constant over the volume of the device. The spectral current responsivity is equal to Ri ˆ

kg qg; hc

…1†

where k is the wavelength, h is PlanckÕs constant, c is the light velocity, and q is the electron charge. Assuming that the current gains for photocurrent and noise current are the same, the current noise due to generation and recombination processes is [5,6] In2 ˆ 2…G ‡ R†Ae t Dfq2 g2 ;

where G and R are the generation and recombination rates, Df is the frequency band, and t is the thickness of the detector. Detectivity D is the main parameter characterising normalised signal to noise performance of detectors and can be de®ned as D ˆ

Ri …Ao Df † In

1=2

:

According to Eqs. (1)±(3),  1=2 k Ao g‰2…G ‡ R†tŠÿ1=2 : D ˆ hc Ae

…3†

…4†

For a given wavelength and operating temperature, the highest performance can be obtained by 1=2 maximising g=‰t…G ‡ R†Š . This means that a high quantum eciency must be obtained with a thin device. In further considerations, we assume that Ao =Ae ˆ 1. Assuming a single pass of the radiation and negligible frontside and backside re¯ection coe€cients, the quantum eciency and detectivity are g ˆ 1 ÿ exp …ÿat†; D ˆ

Fig. 1. Schematic of 3-D heterostructure photodetector (after Ref. [4]).

…2†

k ÿ1=2 …1 ÿ exp … ÿ at††‰2…G ‡ R†tŠ ; hc

…5† …6†

where a is the absorption coecient. The highest detectivity can be obtained for t ˆ 1:26=a for which …1 ÿ exp …at††t1=2 achieves a maximum value of 0:62a1=2 [6]. This thickness is the best compromise between the requirements of high quantum eciency and low thermal generation. In this optimum case g ˆ 0:716 and detectivity is equal:  1=2 k a : …7† D ˆ 0:45 hc G ‡ R

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

To achieve a high performance, the thermal generation must be suppressed to possibly the lowest level. This is usually done by cryogenic cooling of the detector. For practical purposes, the ideal situation occurs when the thermal generation is reduced below the optical generation. At equilibrium, the generation and recombination rates are equal, and we have [6] k  a 1=2 : …8† D ˆ 0:31 hc G The ratio of absorption coecient to the thermal generation rate, a=G, is the fundamental ®gure of merit of any material for IR photodetectors which directly determines the detectivity limits of the devices [6]. An optimised photodetector of any type should provide the highest ratio of the optical-to-thermal generation rates. Any potential material should be compared on this basis. IR photodetectors are traditionally related to photoconductive and photovoltaic detectors based on the principle by which optically generated carriers are detected as a change in voltage or current across the element. The recent advances in heterostructure devices such as development of the heterojunction photoconductors, double layer heterostructure photodiodes and the introduction of non-equilibrium modes of operation make this distinction not so clear. An optimised photodetector of any type should consist of the following regions (Fig. 1) [4]: · Lightly doped narrow gap semiconductor region, which acts as an absorber of IR radiation. Its band gap, doping and geometry should be selected. · Electric contacts to the narrow gap region which sense optically generated charge carriers. Contacts should not contribute to the dark current of the device. · Passivation of the narrow gap region. The surfaces of the absorber regions must be insulated from the ambient by a material that also does not contribute to the generation of carriers. In addition, the carriers which are optically generated in the absorber are kept away from surfaces, where recombination can reduce the quantum eciency. For the best sensitivity, the frontside face should perfectly transmit IR radiation.

215

· Refractive, re¯ective or di€ractive concentrator of IR radiation. · Backside mirror for double pass of IR radiation. In practice, such a device can be obtained using three-dimensional gap and doping engineering, with the narrow gap absorber buried in a wide gap semiconductor. The undoped wide gap material can be used as a window, and/or a concentrator of incoming radiation. Doped n- and p-type semiconductors are used for contacts. Thermal generation in the contact regions is virtually eliminated by making them wide gap. The generation±recombination processes are associated with the predominant recombination mechanisms. There are three important generation and recombination mechanisms: Shockley±Read, radiative, and Auger mechanisms. The Shockley± Read mechanism occurs via lattice defects and impurity energy levels within the forbidden energy gap. This mechanism set can be controlled by the procedure used to grow the material; consequently, the Shockley±Read process is not a fundamental limit to the performance of the detector. The radiative generation±recombination and Auger mechanisms are fundamental band-to-band processes, which are determined by the electronic band structure of the semiconductor. It appears that Auger mechanisms dominate generation and recombination processes in high-quality InSb-like narrow gap semiconductors. The generation rate due to the Auger 1 and Auger 7 processes can be described as [7,8]   n p 1 p n‡ ; …9† GA ˆ i ‡ i ˆ i c 2sA1 2sA7 2sA1 where siA1 and siA7 are the intrinsic Auger 1 and Auger 7 recombination times. Then Auger dominated detectivity is equal to (see Eq. (4))  1=2 k g siA1  ; …10† D ˆ 1=2 2 hc t1=2 n ‡ p=c where c ˆ siA7 =siA1 . As the resulting Auger generation rate achieves its minimum for p ˆ c1=2 ni , it leads to an important conclusion about optimum doping. The optimum performance of Auger limited detectors can be achieved with a lightly p-type doping with whole concentration p ˆ c1=2 ni .

216

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Assuming that the saturation dark current Is of a photodiode is only due to thermal generation in the base layer and that its thickness is low compared to the di€usion length, we have [7,8] Is ˆ qGAt;

…11†

where G is the generation rate in the base layer. Because the zero bias resistance-area product is Ro A ˆ

kTA ; qIs

…12†

kT : q2 Gt

…13†

so Ro A ˆ

Taking into account the Auger 7 mechanism in extrinsic p-type region of n‡ -on-p photodiode, we obtain Ro A ˆ

2kT siA7 q2 N a t

…14†

and the same equation for p-on-n photodiode Ro A ˆ

2kT siA1 ; q2 N d t

…15†

where Na and Nd are the acceptor and donor concentrations in the base region, respectively.

3. HgCdTe ternary alloy for infrared detectors During the past four decades, mercury cadmium telluride (HgCdTe) has become the most important semiconductor for the middle and long wavelength (k ˆ 3±30 lm) IR photodetectors [1,2,4,9±15]. The short wavelength region has been dominated by III±V compounds (InGaAs, InAsSb, InGaSb). There have been numerous attempts to replace HgCdTe with alternative materials. At present, several other variable gap alloy systems are known including closely related mercury alloys (HgZnTe, HgMnTe), lead tin tellurides and selenides, InAsSb, III±VI compounds with thallium and bismuth, free-carrier detectors and low dimensional solids [14±18].

The main motivations for the numerous attempts to replace HgCdTe are technological problems of this material. One of them is a weak Hg±Te bond, which results in bulk, surface and interface instabilities. Uniformity and yield are still issues. Nevertheless, HgCdTe remains the leading semiconductor for IR detectors. The most important reasons for this are the following: · None of the new materials o€ers fundamental advantages over HgCdTe. While the ®gure of merit, …a=G†1=2 , of various narrow gap semiconductors seems to be very close to that of HgCdTe, the free carrier detectors and GaAs/ AlGaAs superlattice devices have several orders of magnitude smaller a=G. · HgCdTe exhibits extreme ¯exibility. It can be tailored for optimised detection at any region of IR spectrum, dual and multicolour devices can be easily constructed. · The present development of IR photodetectors has been dominated by complex band gap heterostructures. Among various variable band gap semiconductor alloys, HgCdTe is the only one material covering the whole IR spectral range having nearly the same lattice parameter. The di€erence of lattice parameter between CdTe …Eg ˆ 1:5 eV† and Hg0:8 Cd0:2 Te …Eg ˆ 0:1 eV† is 0.2%. Replacing a small fraction of Cd with Zn or Te with Se can compensate the residual lattice mismatch. The independence of lattice parameter on composition is a major advantage of HgCdTe over many other materials. Actually, heterostructures do not o€er any adherent fundamental advantages over homostructures for the conventional equilibrium mode devices. The fundamental limits for performance of IR detectors are imposed by unavoidable physics of optical and thermal generation in the narrow gap base region of a photodetector. Nevertheless, heterojunctions are helpful in achieving high performance in practice. For example, the narrow gap HgCdTe which absorbs infrared radiation can be buried encapsulated in wider gap HgCdTe, preventing instabilities due to the weak Hg±Te bonds. In addition, heterostructures can be used for non-equilibrium device the potential performance of which can be much higher than that of the conventional ones [12].

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Epitaxy is the preferable technique to obtain device-quality HgCdTe epilayers for IR devices. The epitaxial techniques o€er, in comparison with bulk growth techniques, the possibility to grow large area epilayers and sophisticated layered structures with abrupt and complex composition and doping pro®les, which can be con®gured to improve the performance of photodetectors. The growth is performed at low temperatures, which makes it possible to reduce the native defect density. Due to the low mercury pressures, there is no need for thick-walled ampules and the growth can be carried out in reusable production-type growth systems. The as-grown epilayers can be lowtemperature annealed in situ. Epitaxial growth of the HgCdTe detector array on a Si substrate, rather than CdZnTe, has emerged as a particularly promising approach to scale up wafer dimensions and achieve a coste€ective number of array die from each processed wafer. In addition to the potential for increasing wafer size from the current 30 cm2 for CdZnTe substrates to 125 cm2 for Si substrates, the growth of HgCdTe FPAs on Si substrates o€ers other compelling advantages such as creation of a thermal-expansion matched hybrid structure, superior substrate mechanical strength and ¯atness, elimination of impurity out di€usion from the substrate, and compatibility with automated wafer processing and handling methodologies. Among various epitaxial techniques, liquid phase epitaxy (LPE) is the most mature method enabling growth of device-quality homogeneous layers and multi-layered structures. LPE growth must be performed at relatively high growth temperature with attendant interdi€usion and resulting graded interfaces. The recent e€orts are mostly on low growth temperature techniques: metalorganic chemical vapour deposition (MOCVD) and molecular beam epitaxy (MBE). MOCVD is a non-equilibrium method, which appears to be the most promising for future large-scale and low-cost production of epilayers. The important advantage of this method is the reduced growth temperature and the ability to modify the growth conditions during growth to obtain the required band-gap and doping pro®les.

217

Intensive studies are currently under way on MBE. This technique o€ers unique capabilities in material and device engineering, including the lowest growth temperature, superlattice growth and potential for the most sophisticated composition and doping pro®les. The growth temperature is less than 200°C for MBE but around 350°C for MOCVD, making it more dicult to control the p-type doping in the MOCVD due to the formation of Hg vacancies at higher growth temperatures. The main drawback of both technologies is a high cost of equipment and maintenance. This prevents a wider spread of the methods. 4. Double layer heterojunction photodiodes The realisation of HgCdTe heterostructure photodiodes is based on P‡ -on-n structure (symbol ``‡'' denotes strong doping, the capital letters means the materials with larger band-gap energy). In such diodes, the lightly doped narrow gap absorbing region (``base'' of the photodiode) determines the dark current and photocurrent. In these photodiodes, the base n-type layers are sandwiched between CdZnTe substrate and highdoped, wider-gap regions. Due to backside illumination (through CdZnTe substrate) and internal electric ®elds (which are ``blocking'' for minority carriers), in¯uence of surface recombination on the photodiodes performance is eliminated. The in¯uence of surface recombination can be also prevented by the use of suitable passivation. Both optical and thermal generations are suppressed in the P‡ -region due to wide gap. Thus the Ro A product of double-layer heterojunction (DLHJ) structures is higher than that for homostructures. The thickness of the base region should be optimised for near unity quantum eciency and a low dark current. This is achieved with a base thickness slightly higher than the inverse absorption coecient for single pass devices: t ˆ 1=a (which is 10 lm) or half of the 1=a for double pass devices (devices supplied with a retrore¯ector). Low doping is bene®cial for low thermal generation and high quantum eciency. As the di€usion length in the absorbing region is typically longer than its thickness, any carriers generated in

218

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

the base region can be collected giving rise to the photocurrent. The baseline detector architectures for short wavelength infrared (SWIR) InGaAs and HgCdTe DLHJ photodiodes are shown in Fig. 2. Fig. 3 shows the schematic band pro®les of the unbiased HgCdTe heterojunction photodiode. Usually, a small gap gradient exists in the base layer, which is either due to interdi€usion or due to variation of conditions during growth. This gradient results in an electric ®eld, which is bene®cial for collection of photogenerated carriers. The gap grading in the cap layer is typically large. It was shown [19±27] that, if the compositional grading extends into the undepleted base region, potential barriers to the ¯ow of the minority carriers generated in the base layers can be formed and result in a very low quantum eciency. The minimum requirement for the formation of carrier collection barriers in HgCdTe heterojunctions is compositional grading, which extends into the neutral base region beyond the edge of the junction space charge region. Once this is present, the severity of the e€ect depends on the height of the barrier, the temperature, and the bias voltage of the device. Compositional di€erences as low as 0.04 can cause barriers in Hg1ÿx Cdx Te [19]. The critical parameters ± composition di€erence across the heterojunction …Dx†, composition gradient, placement of the P±n junc-

Fig. 3. Schematic band diagrams of epitaxial P‡ -on-n HgCdTe heterojunction photodiode.

tion relative to the composition junction, and base dopant concentration …Nd † ± are technologyspeci®c to each material growth process [27]. In the case of InGaAs photodiodes, the starting substrate is n‡ ±InP on which is deposited approximately 1 lm of n‡ ±InP as a bu€er layer. 3±4 lm of the nÿ ±InGaAs active layer is then deposited, followed by a 1 lm nÿ ±InP cap layer. The structure is covered with Si3 N4 . The p-on-n photodiodes are formed by the di€usion of zinc through the InP cap into the active layer. Ohmic contacts were formed by the sintering of a Au/Zn alloy. At this point, the substrate is thinned to approximately 100 lm and a sintered Au/Ge alloy is used as the back ohmic contact.

Fig. 2. Double layer planar heterostructure cross-section for (a) InGaAs and (b) HgCdTe photodiodes.

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

In HgCdTe photodiode technology, n-type base absorbing region is deliberately doped with indium at a level of about …1±3†  1015 cmÿ3 and the p±n junction is formed using arsenic as the dopand at a level of about 1018 cmÿ3 [27±37]. To activate As an acceptor, it must occupy a Te side in the lattice. Full As activation is achieved for annealing temperatures of 300°C or higher, followed by annealing at 250°C in Hg pressure to annihilate Hg vacancies. The junction are also formed by As selective implantation through windows made on a mask of photoresist/ZnS and then di€using the arsenic through the cap layer into the narrow-gap base layer. After implantation, the sample underwent two consecutive annealings, one at about 430°C (to di€use the arsenic into the base layer) for approximately 10 min and the other at 250°C (to annihilate Hg vacancies formed in the HgCdTe lattice during growth and di€usion of arsenic) [36]. The electrical junction is positioned near the metallurgical interface and it is wise to place the junction in the small band-gap layer to avoid deleterious e€ects on the quantum eciency and dark currents. At present, most laboratories use CdTe or CdZnTe (deposited by MBE, MOCVD, sputtering and e-beam evaporation) for photodiode passivation [38,39]. 4.1. Long wavelength infrared photodiodes The dependence of the base region di€usion limited Ro A product on the long wavelength cut-o€ for p‡ -on-n long wavelength IR (LWIR) HgCdTe photodiodes at di€erent temperatures is shown in Fig. 4. This ®gure also includes the experimental data reported by many authors for DLHJ p-on-n structures. The wider band-gap cap layer contributes a negligible amount of thermally generated di€usion current compared with that from an ntype absorber layer at 77 K, and the upper experimental data are situated about a half of an order below ultimate theoretical predictions. With a lowering of the operational temperature of photodiodes, the discrepancy between the theoretical curves and experimental data increases, which is due to additional currents in the junctions (such as tunnelling current or surface leakage current) that are not considered. Photodiodes with

219

Fig. 4. Dependence of the Ro A product on the long wavelength cut-o€ for LWIR p‡ ±n HgCdTe photodiodes at temperatures 6 77 K. The solid lines are calculated assuming that the performance of photodiodes are due to thermal generation governed by the Auger mechanism in the base n-type region of photodiodes with t ˆ 10 lm and Nd ˆ 5  1014 cmÿ3 . The experimental values are taken from di€erent papers marked inside the ®gure.

lower performance usually contain metallurgical defects such as dislocation clusters and loops, pin holes, striations, Te inclusions, and heavy terracing. It should be noticed that the upper experimental data in the very long wavelength range (above 14 lm) at lower temperature (40 K) coincides very well with the theoretical predictions. The best devices continued to be di€usion-current limited by Auger mechanism at zero bias up to 35 K. At 40 K, the measured Ro A is 2  104 X cm2 and the measured cut-o€ wavelength is 17.6 lm. At 35 K, Ro A is 2  105 X cm2 at cut-o€ wavelength 18.1 lm. Also, the performance of photodiodes with cut-o€ wavelengths of 20.3 lm at 40 K is di€usion limited and Ro A products for the diodes reach

220

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

values in the 103 X cm2 range. These are the highest reported values at long cut-o€s wavelengths for any HgCdTe device. Typical results achieved for LPE based p-on-n DLHJ photodiodes include the following [39]: · Ro A values at 78 K which ®t a di€usion-limited mechanism over a cut-o€ wavelength range from 9 to 11.5 lm for material doped n-type at 2  1015 cm3 and with lifetime 500 ns. · when normalised to a cut-o€ wavelength of 10.5 lm, the average Ro A produced under a background photon ¯ux of 2  1012 photons cmÿ2 sÿ1 is 300 X cm2 with a breakdown voltage of 600 mV, · with a f=2 cold shield to establish UB  3  1016 photons cmÿ2 sÿ1 average Ro A  200 X cm2 , · at 40 K, Ro A is typically above the 6  105 X cm2 required for background limited performance at UB  1  1012 photons cmÿ2 sÿ1 , · even the smallest diodes are not surface limited, · these devices also withstand prolonged baking at 150°C, · exceptionally low 1=f noise (knee frequency <0.1 Hz at 50 mV reverse bias) has been observed. Near lattice matched CdZnTe substrates su€er from severe drawbacks such as a lack of large area, high production cost, and more importantly, the di€erence of thermal expansion coecient in CdZnTe substrates and silicon readout integrated circuits as well as interest in large area based IR FPAs …1024  1024†, have resulted in CdZnTe substrate application limitations. The use of Si substrates is very attractive in IRFPA technology not only because it is less expensive and available in large area wafers but also because in an FPA structure, the coupling of the Si substrates with Si readout circuitry allows the fabrication of very large arrays exhibiting long-term thermal cycle reliability. Despite the large lattice mismatch (19%) between CdTe and Si, MBE has been successfully used for the heteroepitaxial growth of CdTe on Si. Since 1989, Santa Barbara Research Center has successfully utilised ``in®nite-melt'' vertical LPE technology from Hg-rich solution to grow highquality epitaxial HgCdTe on the Si-based alternative substrates for the fabrication of p-on-n

DLHJ detectors for high-performance MWIR FPAs [40]. During the past several years, progress has been made in the growth of MBE and MOCVD CdTe on silicon substrates as well as MBE- and MOCVD-HgCdTe growth on these alternative substrates [41±46]. Di€erent procedures have been used to fabricate composite substrates. de Lyon et al. [43] have fabricated bu€er layer structures consisting of a 1 lm thick layer of ZnTe followed by 8 lm thick CdTe grown at a rate of 1.0 lm hÿ1 at Si(1 1 2) substrate temperature of 270°C. A more complex procedure has been used by Wijewarnasuriya et al. [45]. Following special chemical treatment and predeposition of few SiTe2 monolayers (created by exposing the surface to Te ¯ux), a thin ZnTe layer was grown at 220°C followed by 10 min annealing under CdTe and Te ¯ux at 380°C. Finally, CdTe(2 1 1)B layer was grown at a temperature of 300°C. Using optimised growth condition for Si(2 1 1)B substrates, CdTe(2 1 1)B layers with EPD of 105 ±106 cmÿ2 range could be obtained. Fig. 5 compares the results of LWIR arrays fabricated on MBE-grown CdZnTe/Si and MOCVD-grown CdZnTe/GaAs/Si with a historical trendlines of arrays fabricated on bulk CdZnTe. Each data point represents the array average Ro A product measured at f=2 FOV (300 K) background

Fig. 5. Comparison of the results of LWIR HgCdTe arrays fabricated on MBE-grown CdZnTe/Si and MOCVD-grown CdZnTe/GaAs/Si with a historical trendlines of arrays fabricated on bulk CdZnTe. Each data point represents the array average Ro A product measured at f=2 FOV (300 K) background at a temperature of 78 K (after Ref. [41]).

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

at a temperature of 78 K. LWIR HgCdTe DLHJ were grown by vertical LPE from in®nite melt Hgrich solutions. The n-type base layer was doped with indium and the wider band-gap layer was doped with arsenic. The two continuous curves shown in Fig. 5 were calculated to indicate di€usion-limited behaviour and are only shown as a guide to the eye. We can see that there is no signi®cant di€erence between arrays fabricated on either CdZnTe/Si or CdZnTe/GaAs/Si and the results on these Si-substrates are comparable with results on bulk CdZnTe substrates at 78 K. To improve reverse-bias characteristics at 78 K and improve detector performance at lower temperatures in comparison with bulk CdZnTe substrates, a further reduction in the dislocation density for HgCdTe-grown on Si-based substrates is needed. For low-background applications, the HgCdTe photodiodes are operated at 40 K. It is well recognised that with a lowering of the operation temperature of photodiodes, the discrepancy between the theoretical curves and experimental data increases. The Ro A product distribution for p-on-n DLHJ devices at this temperature spans a wide range of several orders of magnitude. Chen et al. [47] carried out a detailed analysis of the wide distribution of the Ro values. Fig. 6 shows the cumulative distribution functions of Ro values at

Fig. 6. Detailed analysis separates the cumulative distribution function of Ro A values of LWIR p-on-n HgCdTe photodiodes (fabricated by LPE) into three regions: good diodes, diodes a€ected by point defects, and diodes a€ected by metallurgical defects (after Ref. [47]).

221

40 K from devices with a cut-o€ wavelength between 9.4 and 10.5 lm. It is clear that while some devices exhibit a fair operability with Ro values spanning only two orders of magnitude, other devices show a poor operability with Ro values spanning more than ®ve to six orders of magnitude. Lower performance, with Ro values below 7  106 X at 40 K, usually contained gross metallurgical defects such as dislocation clusters and loops, pin holes, striations, Te inclusions, and heavy terracing. However, diodes with Ro values between 7  106 and 1  109 X at 40 K contained no visible defects (Hg interstitials and vacancies). Dislocations, twins and subgrain boundaries in LWIR n‡ ±p HgCdTe photodiodes primarily produce bias dependent dark current [48], while Te precipitation and associated dislocation multiplication produces bias dependent noise [49]. InAs/Ga1ÿx Inx Sb strained layer superlattices (SLSs) have also been proposed for IR detector applications in the 8±14 lm region [50,51]. It has been suggested that this material system can have some advantages over bulk HgCdTe, including lower leakage currents and greater uniformity. Long wavelength response in these SLs arises due to a type II band alignment and internal strain which lowers the conduction band minimum of InAs and raises the heavy-hole band in Ga1ÿx Inx Sb by the deformation potential e€ect. This reduced band gap is advantageous because longer cut-o€ wavelengths can be obtained with reduced layer thickness in the strained SL, leading to optical absorption coecient comparable to that of HgCdTe. High performance InAs/GaInSb SL photovoltaic operation is predicted by the theoretical promise of longer intrinsic lifetimes due to the suppression of Auger recombination mechanism [52]. Fig. 7 compares the theoretical temperature dependence of detectivity of two types of HgCdTe photodiodes and InAs/InGaSb SL photovoltaic detectors operating at 11 lm. The curves calcu InAs/15 A  lated for two InAs/GaInSb SLs (39.8 A  InAs/25 A  Ga0:75 In0:25 Sb) Ga0:6 In0:4 Sb and 41 A are taken from Ref. [53]. As Fig. 7 shows, the ultimate detectivity of HgCdTe photodiodes with optimally doped base region are comparable with that of InAs/InGaSb SLs in the temperature

222

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 7. Calculated detectivity of three types of HgCdTe photodiodes and InAs/InGaSb SL photovoltaic detectors operated at 11 lm as a function of temperature. The base-region limited detectivities are calculated for HgCdTe photodiodes assuming g ˆ 1, t ˆ 10 lm, Na ˆ 5  1015 cmÿ3 for n‡ ±p structures; g ˆ 1, t ˆ 10 lm, Nd ˆ 3  1014 cmÿ3 for p‡ ±n structures; and g ˆ 1, t ˆ 10 lm for photodiode with optimal doping in the base region. The curves calculated for two InAs/InGaSb  InAs/15 A  In0:4 Ga0:6 Sb and 41 A  InAs/25 A  SLs (39.8 A In0:25 Ga0:75 Sb) are taken from Ref. [53] (after Refs. [54,55]).

region between 300 and 77 K [54,55]. High performance of InAs/GaInSb SLSs detectors (also at higher operating temperatures) is a result of long carrier lifetimes caused by a the splitting of the light-hole and heavy-hole bands. Recombination in SLSs occurs by a relatively slow Auger recombination process. Fundamental material issues still remain an obstacle in realising high performance SLS FPAs. As grown material has exhibited a high residual ntype doping. Non-optimised carrier lifetimes have been observed [56] and at desirably low carrier concentrations the SLS material have a Shockley± Read limited lifetime of <10 ns with a minimum obtainable threshold of around 1016 cmÿ3 for the background doping [57]. We close our discussion concerning the fabrication of InAs/GaInSb SLSs by considering only some aspects of the hetero-epitaxial growth of this system. More information can be found, e.g. in MailhiotÕs paper [51]. The substrate GaSb is a convenient material on which to grow InAs/

GaInSb superlattices. However, commercial GaSb substrates are available only as unintentionally ptype, or doped with Te to relatively high n-type levels and exhibit extrinsic or free-carrier absorption extending into IR regions. Despite the relatively low absorption coecients, GaSb substrates require thinning the thickness below 25 lm in order to transmit appreciable IR radiation [57]. This predicament complicates the fabrication of hybrid FPAs. The ®rst InAs/GaInSb SLS photodiodes with photoresponse out to 10.6 lm, have been presented by Johnson et al. [58]. The detectors consist of double heterojunctions (DH) of the SLS with ntype and p-type GaSb grown on GaSb susbstrates. Fig. 8(a) shows the results of calculations of the band structure applied to a device model consisting of P-type (2  1017 cmÿ3 ) GaSb contact layer, an n-type (2  1016 cmÿ3 ) active region corre InAs/16 A  Ga0:65 In0:35 Sb SLS sponding to a 39 A 17 ÿ3 and an N-type (8  10 cm ) GaSb contact layer. Fig. 8(b) shows a schematic illustration of a DH SLS photodiode grown by MBE. The transport barriers formed at the heterointerfaces have shown to be problematic [58]. It has been theoretically estimated that n-on-p geometry to be desirable based on favourable Auger lifetime and superior minority carrier transport properties associated with p-type SLS [57]. Fig. 9 compares the Ro A values of InAs/ GaInSb SLS and HgCdTe photodiodes in the long wavelength spectral range [57±61]. The upper line denotes the theoretical di€usion limited performance corresponding to Auger 7 limitation in p-type HgCdTe material. As can be seen in the

 Fig. 8. DH SLS photodiode: (a) band diagram of the 39 A  Ga0:65 In0:35 Sb SLS, (b) schematic of a mesa photoInAs/16 A diode (after Ref. [58]).

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 9. Dependence of the Ro A product of InAs/GaInSb SLS photodiodes on cut-o€ wavelength compared to theoretical and experimental trendlines for comparable HgCdTe photodiodes at 78 K (after Ref. [57]).

®gure, the most recent photodiode results for SLS devices rival that of practical HgCdTe devices, indicating substantial improvement has been achieved in SLS detector development. However, signi®cant obstacles in material growth and device fabrication need to be addressed before its full potential can be realised. 4.2. Middle wavelength infrared photodiodes Middle wavelength infrared (MWIR) HgCdTe photodiodes were the ®rst to be developed, and many mature technologies have been used to demonstrate FPAs [62,63]. Rockwell has hence developed the capability for fabricating large MWIR HgCdTe FPAs by epitaxially growing HgCdTe on sapphire (PACE-1 process). In his process a CdTe layer is grown via MOCVD on the sapphire, and next HgCdTe is grown on the CdTe bu€er via LPE. The junctions are formed by boron ion implantation and thermal annealing. Planar and mesa junctions are used, depending on the FPA speci®cations, and are passivated with a ZnS or CdTe ®lm. The detector array is backside illuminated through the sapphire substrate, which transmits to 6.5 lm for a 7 mil thickness. PACE-1 process currently provides intrinsic detector arrays with BLIP performance and satisfactory yield. The sapphire substrate o€ers several attributes. Its thermal coecient of expansion matches well to the alumina chip carrier, thereby greatly improv-

223

ing large hybrid reliability. The large substrate also reduces detector cost by increasing throughput, enabling batch processing and reducing breakage. The 3 in wafers, currently in production, enable populating each wafer with either ®ve 18.5 lm pixel 1024  1024 arrays (in addition to four 40 lm 256  256 arrays), eight 27 lm 640  480 arrays, or 21 256  256 arrays. Since 1989, SBRC has successfully utilised ``in®nite-melt'' vertical LPE technology from Hgrich solution to grow high-quality epitaxial HgCdTe on the Si-based alternative substrates for the fabrication of p-on-n DLHJ detectors for highperformance MWIR FPAs. Tung et al. [40] reported large, up to 480  640, MWIR FPAs, grown on Si-based alternative substrates. Fig. 10 compares the Ro A product for two PACE MWIR HgCdTe layers with cut-o€ wavelengths at 78 K of 5.11 and 4.65 lm versus temperature. Also compared is the theoretical Ro A performance for 5 lm p-on-n HgCdTe photodiode; this level is readily achieved with MBE MWIR DLHJ HgCdTe/CdZnTe material [64]. Using a buried planar heterostructure produces devices whose performance is less critical of passivation as compared to conventional mesa technology because the junction interface is buried. The planar p-on-n photodiodes are formed by selective pocket di€usion of arsenic (a p-type dopant), which is deposited by ion implantation on the wide band-gap cap layer.

Fig. 10. Ro A product versus temperature for MWIR HgCdTe photodiodes (after Ref. [64]).

224

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 11. Comparison of 125 K detector performance for MWIR HgCdTe photodiodes grown on Si and CdZnTe by MBE and photodiodes grown on CdZnTe by LPE. Each data point represents an array-median Ro A product measured at 125 K (after Ref. [46]).

The MBE-grown devices on Si and CdZnTe appear to be identical [46]. Fig. 11 presents a comprehensive comparison of the performance of MWIR P‡ -on-n HgCdTe photodiodes on CdZnTe and Si substrates for cut-o€ wavelengths ranging from 3.5 to 5 lm. The various data points are median values for mini-arrays included in test structures for each processed wafer. The devices with highest performance are processed from MBE-grown epilayers on bulk CdZnTe substrates. The shorter cut-o€ devices (with kc  3 lm) are di€usion-limited down to at least 125 K. The devices with longer cut-o€ wavelength (with kc  5 lm) appear to be di€usion-limited down to approximately 110 K. Below this temperature, the experimental data obscure the probable onset of generation±recombination and/or tunnelling current limitations. 4.3. Short wavelength infrared photodiodes Considerable progress in HgCdTe SWIR hybrid FPAs has been achieved in the last decade. At the beginning the detector arrays were fabricated using an n‡ boron implanted process on p-type HgCdTe layers grown by LPE on CdTe or CdZnTe substrates [62]. Next, the PACE-1 process was adopted to fabricate large, 2.5 lm 1024  1024, FPAs (HAWAII) for IR astronomy [65,66].

Recently, DLPH p-on-n photodiodes in MBE HgCdTe on CdZnTe substrates have been elaborated by As-ion implantation and the p-dopant activation by an open-tube Hg anneal [67]. The highest detector performance is achieved by growing the layers on lattice-matched substrates such as CdZnTe (their properties are least a€ected by threading dislocations). Adding In during growth provided donor doping nominally 1:5  1015 cmÿ3 with a majority carrier mobility of 8000 cm2 Vÿ1 sÿ1 at 77 K. The MBE DLHJ HgCdTe/CdZnTe photodiodes are made using a buried planar heterostructure to produce devices whose performance is less critical of passivation as compared to conventional mesa technology because the junction interface is buried. Typical active layer thickness is 3.5 lm and cap layer thickness is 0.4 lm [67]. The planar p-on-n photodiodes are subsequently form by selective pocket di€usion of arsenic (a p-type dopant), which is deposited by ion implantation on the wide band-gap cap layer. The highest quality SWIR HgCdTe photodiodes has performance in agreement with the radiative limit. It appears, however, that due to photon recycling, an order of magnitude enhancement in the radiative lifetimes over those obtained from the standard van Roosbroeck and Shockley expression is observed in materials like In0:53 Ga0:47 As lattice-matched to InP substrates [68]. The same situation can be observed in HgCdTe ternary alloys [69±71]. A consequence of enhancement in the radiative lifetime is higher ultimate performance of the photodiodes. Assuming the same situation for SWIR HgCdTe photodiodes, Rogalski and Ciupa have reconsidered the ultimate performance of p-on-n HgCdTe photodiodes and compared theoretical predictions with attainable experimental data [72±76]. In calculations it was assumed that typical doping concentrations in the active region of high quality photodiodes is 3  1015 cmÿ3 . Due to the similar band structure of InGaAs and HgCdTe ternary alloys, the ultimate fundamental performance of both types of photodiodes are similar in the wavelength range 1:5 < k < 3:7 lm [68]. Figs. 12 and 13 compare the ultimate performance of n-type base InGaAs and HgCdTe photodiodes with available experimental data.

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 12. The dependence of e€ective Ro A product on the long wavelength cut-o€ for SWIR InGaAs photodiodes at room temperature. The calculations are performed assuming that the performance of photodiodes is due to fundamental generationrecombination processes in the base n-type region of photodiodes with t ˆ 5 lm and Na ˆ 3  1015 cmÿ3 . The experimental values are taken from Ref. [72] (), Ref. [73] (), Ref. [74] ( ), and Ref. [75] (n) (after Ref. [68]).

InGaAs photodiodes have shown high device performance close to theoretical limits for materials whose composition is nearly matched to that of InP (1.7 lm cut-o€ wavelength) and InAs (3.6 lm cut-o€ wavelength). However, their performance decreases rapidly at intermediate wavelengths due to mismatch-induced defects with the substrate. On the other hand, SWIR HgCdTe photodiodes have good performance over a wider range of wavelengths. This is due to lattice match of active base photodiode layers with CdZnTe substrate and consequently lower in¯uence of induced defects at interface on photodiode leakage current.

5. Dual-band HgCdTe detectors As IR technology continues to advance, there is a growing demand for multispectral detectors for

225

Fig. 13. The dependence of e€ective Ro A product on the long wavelength cut-o€ for SWIR HgCdTe photodiodes at room temperature. The calculations are performed assuming that the performance of photodiodes is due to fundamental generationrecombination processes in the base n-type region of photodiodes with t ˆ 5 lm and Na ˆ 3  1015 cmÿ3 . The experimental values are taken from Ref. [67] (), Ref. [72] ( ), and Ref. [76] (n) (after Ref. [68]).

advanced IR systems with better target discrimination and identi®cation. Systems that gather data in separate IR spectral bands can discriminate both absolute temperature and unique signatures of objects in the scene. By providing this new dimension of contrast, multiband detection also enables advanced colour processing algorithms to further improve sensitivity above that of singlecolour devices. This is extremely important for the process of identifying temperature di€erences between missile targets, war heads, and decoys. Multispectral IR FPAs can also play many important roles in Earth and planetary remote sensing, astronomy, etc. Currently, multispectral systems rely on cumbersome imaging techniques that either disperse the optical signal across multiple IR FPAs or use a ®lter wheel to spectrally discriminate the image focused on single FPA. These systems contain beam-splitters, lenses, and band-pass ®lters in the optical path to focus the images onto separate

226

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

FPAs responding to di€erent IR bands. Also, complex alignment is required to map the multispectral image pixel for pixel. Consequently, these approaches are expensive in terms of size, complexity, and cooling requirements. At present, considerable e€ort is directed towards fabricating a single FPA with multicolour capability to eliminate the spatial alignment and temporal registration problems that exist whenever separate arrays are used. This ``integration'' of multispectral capability a€ords sensitivity to different IR bands within each and every unit cell of the array so that temporal and spatial co-registration between each spectral ®eld now occurs on the pixel level. This approach o€ers multicolour advantages without the complexity of multiple FPA systems, thereby o€ering signi®cant reduction of weight and power consumption in a simpler, more reliable and less costly package. This can be implemented as a capability upgrade to existing single colour systems with only minor modi®cations to FPA control and signal processing electronics, since power and space requirements for multispectral FPAs are identical to single-colour FPAs. It is expected that beyond ®ve years, as the two-colour array technology in demonstration currently moves into production, there will be demonstrations of three- and fourcolour capabilities squeezed into one pixel. Four colours may be about the limiting number of bands that can be stacked in a single pixel. For applications requiring greater spectral decomposition, alternative approaches are being developed [77,78]. Both HgCdTe photodiodes and quantum well IR photodetectors (QWIPs) o€er multicolour capability in the MWIR and LWIR range. Each of these technologies has its advantages and disadvantages. QWIP technology is based on the well developed A3 B5 material system which has a large industrial base with a number of military and commercial applications. The HgCdTe material system is only used for detector applications. Therefore, QWIPs are easier to fabricate with high yield, high operability, good uniformity and lower cost. On the other hand, HgCdTe FPAs have higher quantum eciency, higher operating temperature and potential for the highest perfor-

mance. A more detailed comparison of both technologies has been recently given by Tidrow and Rogalski [79,80]. They compared the technical merits of two IR detector arrays technologies: photovoltaic HgCdTe and QWIPs. It was clearly shown that LWIR QWIP cannot compete with HgCdTe photodiode as the single device especially at higher operational temperatures (>70 K) due to fundamental limitations associated with inter-subband transitions. However, the advantage of HgCdTe is less distinct in temperature range below 50 K due to problems involved in HgCdTe material (p-type doping, Shockley±Read recombination, trap-assisted tunnelling, surface and interface instabilities). Even though the QWIP is a photoconductor, several of its properties such as high impedance, fast response time, long integration time, and low power consumption, comply well with the fabrication requirements of large FPAs. Due to the high material quality at low temperature, QWIP has potential advantages over HgCdTe for LWIR (VLWIR) FPA applications in terms of the array size, uniformity, yield and cost of the systems. Considerable progress has been recently demonstrated by research groups at Hughes Research Laboratory [81±84] and Lockheed Martin [85±88] in multispectral HgCdTe detectors employing MBE and MOCVD for the growth of a variety of devices. Devices for the sequential and simultaneous detection of two closely spaced sub-bands in the MWIR and LWIR radiation have been demonstrated. This section reviews recent progress made in two-colour HgCdTe detector technology from the IR industrial community. It is a culmination of recent progress in critical technologies necessary for multicolour detector development. 5.1. Sequential and simultaneous operation The unit cell of an integrated two-colour FPAs consists of two co-located detectors, each sensitive to a di€erent spectral band. Radiation for both bands is incident on the shorter band detector, with the longer wave radiation passing through to the second detector. This device architecture is realised simply by placing a longer wavelength

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

HgCdTe photodiode behind shorter wavelength photodiode. Back-to-back photodiode two-colour detectors were ®rst implemented using quaternary III±V alloy …Gax In1ÿx Asy P1ÿy † absorbing layers in a lattice matched InP structure sensitive to two di€erent SWIR bands [89]. Integrated two-colour HgCdTe technology has been developed for nearly a decade with a steady progression having a wide variety of pixel size (30±61 lm), array formats …64  64 up to 320  240† and spectral-band sensitivity (MWIR/MWIR, MWIR/LWIR and LWIR/LWIR) [84]. Following the successful demonstration of multispectral detectors in LPEgrown HgCdTe devices [90], the MBE and MOCVD techniques have been used for the growth of a variety of multispectral detectors. The two-colour detector arrays are based upon an n±P±N HgCdTe triple layer heterojunction (TLHJ) design. The TLHJ detectors consist of back-to-back photovoltaic p±n junctions. Vertically stacking the two p±n junctions permits incorporation of both detectors into a single pixel. Both sequential mode and simultaneous mode detectors are fabricated from the multi-layer materials. The mode of detection is determined by the fabrication process. Figs. 14 and 15 show the elements of arrays of two-colour photovoltaic unit cells in both modes. The simultaneous mode re-

227

quires bias contact to the cap layer (Fig. 15), while the sequential mode does not (Fig. 14). The sequential-mode detector has a single indium bump per unit cell that permits sequential bias-selectivity of the spectral bands associated with operating tandem photodiodes. The simultaneous mode detector employs an additional electrical contact to the shared-type centre layer so that each junction can be accessed independently with both signal channels integrated simultaneously. The two indium bumps per unit cell required for the simultaneous mode detectors can be fabricated in relatively small unit cells with high optical ®ll factor. Thus, two-colour HgCdTe FPAs present considerable challenges in three technology arrays. First, a versatile multilayer growth capability is needed to form heterojunctions. Second, fairly sophisticated array processing technology is needed to make tight-geometry features, such as two bumps per unit cell and insulated over-theedge contact metallizations, in unit cells as small as 40  40 lm2 . Third, the silicon readout integrated circuit (ROIC) chip now requires two input circuits per unit cell. The distinguishing feature of the sequential approach is that the p-type cap layer is not contacted. Elimination of the contact has several key advantages [90]:

Fig. 14. Schematic cross-section of integrated photovoltaic two-colour detectors in an n±P±N layer structure for sequential operating mode.

228

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 15. Schematic cross-section of integrated photovoltaic two-colour detectors in an n±P±N layer structure for simultaneous operating mode.

· Only one in-cell indium contact is required, giving a unit cell the simplicity of a single-colour array. · Only one readout per unit cell is required, thereby providing space for higher performance readouts. · The simple structure provides smaller and more producible unit cells (<40 lm). · Near 100% ®ll factor can be achieved in both colours (there are no compound features to disrupt total internal re¯ection of the LWIR signal and no second-plane circuitry to obscure incident radiation). · Each detector is precisely co-located, since no part of the LWIR detector must be given up, to form a cap layer contact. Critical step in device formation is connected with in situ doped p-type As-doped layer with good structural and electrical properties to prevent internal gain from generating spectral crosstalk. The band-gap engineering e€ort consists of increasing the CdTe mole fraction and the e€ective thickness of the p-type layer to suppress out-o€band carriers from being collected at the terminal. The problems with the bias selectable device are the following: its construction does not allow independent selection of the optimum bias voltage for each photodiode, and there can be substantial medium wavelength (MW) crosstalk in the long

wavelength (LW) detector. To overcome the problems of the bias-selectable device, independently accessed back-to-back photodiode dual-band detectors have been proposed. An implementation of the simultaneous mode using a second indium bump in the unit cell is shown in Fig. 16. The mesa shape has become more complicated to provide access to the cap layer for the third contact. Internal gain is very e€ectively suppressed through proper bias of each diode, easing the design and growth emphasis on band-gap engineering. The

Fig. 16. SEM photo of a 64  64 two-colour HgCdTe detector array with 75  75 lm2 unit cells (after Ref. [87]).

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

most important distinction is the requirement of a second readout circuit in each unit cell. Longwave band ®ll factor is reduced from that of the midwave, as some junction area is sacri®ced to provide contact to the buried cap layer, and spatial coincidence is altered. The di€erence between sequential and simultaneous operation becomes to some extent indistinct when two widely separated spectral bands are used, such as the 3±5 and 10±12 lm bands. Photon ¯uxes in the longer wavelength band are generally much higher than in the shorter wavelength band, requiring a signi®cantly shorter integration time for the longwave band, and loss of true simultaneity of signal integration can occur. In this situation a smaller LWIR ®ll factor can be a bene®t in reducing background-generated charge. It should be noticed, that two-colour detector arrays are also connected to the ROIC by vias (Fig. 17) and not by commonly used indium bumps. This technique has been successfully used by Raytheon TI systems for their single colour IR FPAs. Reine and co-workers [85±87] have proposed a novel simultaneous MWIR/LWIR dual-band HgCdTe detector fabricated from a P±n±N±P layers grown in situ by the interdi€used multilayer process (IMP) MOCVD onto lattice-matched CdZnTe substrates. Fig. 18 shows a cross-section of the independently accessed back-to-back photodiode dual-band detector. The LW photodiode is a P-on-n heterojunction grown directly on top of the MW photodiode, which is an n-on-P heterojunction. A thin n-type compositional barrier layer is placed between the MW and LW absorber layers. This barrier layer forms isotype n±N hetero-

229

Fig. 18. Cross-section and energy band pro®le of the independently accessed back-to-back HgCdTe photodiode dualband detector (after Ref. [87]).

junction at the interface, which prevents MW photocarriers from di€using into the LW absorber layer and also prevents LW photocarriers from di€using into the MW absorber layer. In dual-band IR hybrid FPAs with the above detectors, one bump contacts only the p-type region of the LW photodiode; the other bump contacts the n-type region of the LW photodiode, and therefore also the n-type region of the MW photodiode through an over-the-edge metallization. This way the LW photodiode is accessed directly through the two bumps, while the MW photodiode is accessed through the contact to the n-type region and the array ground. These dual-band devices showed that the MW N-on-P heterojunction photodiodes (in which the absorber layers are the p-type layer) have better performance (good Ro A product, signi®cantly improved quantum ef®ciency of 65±80% without anti-re¯ection coating, and classical spectral response) in comparison with previously fabricated N-on-P homojunctions with p-type absorber layers [85]. 5.2. Technology and characterisation of multilayer heterojunctions

Fig. 17. Two-colour HgCdTe detector connected to the ROIC by vias (after Ref. [88]).

Sharp turn o€ of the detectorÕs special response, a feature which is necessary to achieve low crosstalk in either bands, requires the growth of a homogeneous n-type absorber layer at the predetermined alloy composition. Other detector requirements are high quantum eciency (above 70%) in both bands and high Ro A product. To satisfy these requirements it is necessary to have

230

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

precise control over the placement of the electrical junction, which is determined by the location of the arsenic dopant pro®le with respect to that of In in a heterojunction. Due to the nature of the growth process, the capability to control and tailor the alloy composition and the dopant pro®les are in principle possible by MBE and MOCVD. Two-band HgCdTe detectors are usually fabricated from n±P±N layers grown in situ by MBE or MOCVD onto lattice-matched CdZnTe substrate. Using MBE growth, the precision in the xvalue is about 0:005 [83]. For MWIR layers, a variation in the x-value of 0.005 produces a change in the cut-o€ of 0.01 lm at 77 K. The cut-o€ wavelength of each junction is controlled by varying the composition of the HgCdTe used in the layers. Fig. 19 shows the three distinct compositions corresponding to the MBE grown n±P± N layers observed in the representative SIMS pro®le. The average alloy composition in the 6 lm thick MWIR-1 layer is x ˆ 0:35 with standard deviation of 0.0030 (i.e. standard deviation of <1%) and the corresponding values of the 6 lm thick MWIR-2 is 0:33  0:0031. The dip in the xvalue at the two p=n interfaces seen in Fig. 19 is a result of the change in the alloy composition as a

Fig. 19. SIMS pro®le showing the composition in layers with three distinct MWIR compositions (after Ref. [83]).

Fig. 20. SIMS pro®les for In and As in n±P±N structure (after Ref. [83]).

result of a reduction in the growth temperature for the growth of the p-type layer. This reduction in growth temperature is required to facilitate incorporation of adequate amounts of As acceptor impurities in the p-type layer. Fig. 20 shows SIMS pro®les for In and As in a n±P±N structure. Asgrown layers exhibit constant levels of In in the two n-type absorber layers, and the As pro®le exhibits sharp turn on and turn o€ in the intermediate p-type layer. Furthermore, the In and As pro®les indicate negligible dopant di€usion during growth. One key technical issue in epi-grown of twocolour HgCdTe detectors is high dislocation densities. The commercially available Cd0:96 Zn0:04 Te substrates are lattice-matched to LWIRHg1ÿx Cdx Te with x ˆ 0:22. Hg0:78 Cd0:22 Te grown on Cd0:96 Zn0:04 Te substrates exhibit an average EPD of 5  105 cmÿ2 in as-grown layers. Rajavel et al. [83] have observed that MWIR-HgCdTe …x ˆ 0:30±0:35† layers deposited on Cd0:96 Zn0:04 Te substrates, on average, exhibit an etch-pit density (EPD) that is a factor of 5±10 higher than the LWIR-HgCdTe. The increased EPD in the MWIR-HgCdTe is result of the lattice mismatch with the substrate, which is as small as 0.04%. Fig. 21 shows the near-surface EPD measured in as-

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 21. The near-surface etch-pit density in MWIR/MWIR two-colour detector structures (after Ref. [83]).

grown MWIR/MWIR two-colour detector structure. The layers in each of the two sets were grown consecutively. It should be noticed, however, that the Ro A product of MWIR detectors at 77 K is not a€ected by the dislocation density at levels 5  106 cmÿ2 . The performance of LWIRHgCdTe photodiodes at low temperature (40 K below) is expected to be strongly dependent on the EPD in the absorber layer [91]. For LWIR photodiodes at 78 K, the Ro A product begins to decrease at a dislocation density of approximately 106 cmÿ2 [91,92]. Four-layer P±n±N±P structures have been grown on nominally lattice matched CdZnTe (1 0 0) 4±8° misoriented toward (1 1 1) B. Improvements in MOCVD±IMP were incorporated along with the use of iodine donors and more classical ptype doping with arsenic from DMAAs. The 8 lm thick p-layer was grown ®rst (Fig. 18) with an xvalue of 0.40 and doped with arsenic at …1±3†  1017 cmÿ3 . The 8 lm thick MW n-type absorber layer was grown next and doped with iodine at …2±4†  1015 cmÿ3 . Next, a small compositional barrier has been added at the MW/LW absorber layer interface. This was followed by the n-type LW absorber layer, 8±10 lm thick and doped with iodine at …1±2†  1015 cmÿ3 , on which a 2 lm thick p-type layer, doped with arsenic at

231

Fig. 22. SIMS data for a four layer P±n±N±P structure grown in situ by IMP-MOCVD (after Ref. [87]).

…1±3†  1017 cmÿ3 , was grown. The band gap of the p-type cap layer was wider than that of the ntype absorber layer, with Dx ˆ 0:04. The growth run was terminated with a thin CdTe layer (0.15 lm) to prevent outdi€usion of Hg during the interdi€usion anneal and cooldown. Finally, the HgCdTe layers were annealed under Hg-rich conditions for arsenic activation. In the FPA, the wide gap SW p-type layer is the common contact to all detectors in the array. Fig. 16 shows a scanning electron microphotograph of a section of a 64  64 dual-band HgCdTe detector array. The 23 lm deep MW mesas were de®ned by ECR dray etching. Fig. 22 shows SIMS data for the iodine donors, the arsenic acceptors, and the HgCdTe alloy composition. Note the abruptness of the transitions between the various layers, which illustrates the excellent control of donor and acceptor concentrations and of alloy composition. 5.3. State of the art of two-colour HgCdTe detectors Integrated two-colour detectors have been implemented in a number of variations of structure and material for operation in either sequential and simultaneous mode. Fig. 23 shows examples of

232

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 23. Spectral response curves for two-colour HgCdTe detectors in various dual-band combinations of MWIR and LWIR spectral bands (after Ref. [77]).

spectral response from MWIR/MWIR, MWIR/ LWIR, and LWIR/LWIR two-colour devices. Note that there is a minimal crosstalk between the bands, as the short wavelength band absorbs nearly 100% of the shorter wavelengths. One might question whether the additional processing and complexity of building a two-colour detector structure may adversely a€ect the performance, yield, or operability of two-colour devices. Test structure indicates that the separate photodiodes in a twocolour detector perform exactly as single-colour detectors in terms of achievable Ro A product variation with wavelength at a given temperature. The simultaneous mode two-colour detectors were delineated as mesa isolated structures and contacts were made to the top n-type layer and the intermediate p-type layer. Fill factors of 128  128 MWIR/MWIR FPAs as high as 80% were achieved by using a single mesa structure to accommodate the two indium bump contacts required for each unit cell with 50 lm size. The bottom n-type layer served as the common ground. Fig. 24 shows the uniformity of responsivity for each of the bands. Band 1 …2:5±3:9 lm) had operability of 99.9%, with 23 inoperable pixels. Band 2 …3:9±4:6 lm) had operability of 98.9%, with 193 inoperable pixels. Quantum eciencies of 70% were observed in each band without using an anti-re¯ection coating. The Ro A values for the diodes ranged from 8:25  105 to 1:1  106 X cm2 at f =2 FOV. The NEDT for both bands as a function of temperature is shown in Fig. 25. The camera used for these measurements had a 50 mm, f =2:3 lens. Imagery was acquired at temperatures as high as 180 K with no visible degradation in image quality.

Fig. 24. Responsivity uniformity for (a) 2.5±3.9 lm band and (b) 3.9±4.6 lm band (after Ref. [77]).

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Fig. 25. Noise-equivalent di€erence temperature for a twocolour camera having a 50 mm, f=2.3 lens, as a function of operating temperature (after Ref. [77]).

The experimentally demonstrated 64  64 MW/ LW dual-band MOCVD HgCdTe array is characterised in Table 1. The arrays have a unit cell size of 75  75 lm2 and were hybridised to a dual-band silicon multiplexer readout chip that allowed the MW and LW photocurrents to be integrated simultaneously and independently. The MW and LW average cut-o€ wavelengths at 77 K are in the

233

Fig. 26. Relative spectral response data at 78 K for the MW and LW detectors in one of the dual-band detectors in a 64  64 dual-band FPA (after Ref. [87]).

4.27±4.35 and 10.1±10.5 lm ranges, respectively. Fig. 26 shows the relative spectral response data at 78 K for the MW and LW detectors in one of the dual-band FPA. The MW response is quite small for wavelengths beyond 4.5 lm, indicating low LW-to-MW crosstalk (de®ned as the ratio of the MW photocurrent to the LW photocurrent when only LW radiation is incident on the detector),

Table 1 Summary of performance data for 64  64 dual-band HgCdTe FPA (after Ref. [87]) Parameter

MW

LW

Cut-o€ wavelength (avg [r=l]) (lm) Quantum eciency (avg [r=l]) Detectivity (f=2.9) (median) (cm Hz1=2 /W) NEDT (TSCENE ˆ 295 K, sINT ˆ 2:2 ms) (median) (mK) measured: f=2.9, MW: 3.0±4.35 lm, LW: 4.5±10.1 lm projected: f=2.9, MW: 3.0±4.35 lm, LW: 7.0±10.1 lm Stare eciency Fill factor ˆ AOPT =…75  75 lm2 † Dynamic range (average)(dB) Dynamic resistance RD (average) (X) RD AOPT (average) (X cm2 ) Spectral crosstalk LW …7±11 lm† ! MW MW …4:0 lm† ! LW

4.27 (0.5%) 79% (9%) 4:8  1011

10.1 (2.2%) 67% (29%) 7:1  1010

20 12 87% 94% 77 4:2  108 2:4  104

7.5 6.2 87% 39% 75 2:3  106 51

0.4%

)10%

All data were measured at a temperature of 78 K. Signal response was measured with an MW notch ®lter at 4.0 lm or an LW ®lter 7±11 lm. Average values for all 64  64 elements are stated for all parameters except detectivity and NEDT, for which the median values are quoted.

234

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

consistent with the measured crosstalk value of 0.4%. The MW response at wavelength less than 2.9 lm is suppressed, presumably due to a high recombination rate for SW photocarriers at the interface between the wide-gap p-type window layer and the CdZnTe substrate (Fig. 18). The expected ®ltering of the LW photodiode response spectrum by the MW layer is clearly evident in the sharp increase in LW response at 4.3 lm. The measured LW response spectrum is negative at those wavelengths for which the MW response is high, which is attributed to electrical crosstalk due to ®nite series resistance in the input circuit. A simple small-signal model for this crosstalk mechanism predicts that the crosstalk is given by Rs =…RL ‡ 2Rs †, where RL is the dynamic resistance of the LW diode and Rs is the impedance of the input circuit. The are two approaches for reducing this MW-to-LW electrical crosstalk [87]: increasing the LW photodiode dynamic resistance RL by improved junction quality, and reducing series resistance Rs by replacing the direct injection circuit with, for example, a bu€ered direct injection circuit. These staring dual-band FPAs exhibit high average quantum eciency (MW: 79%; LW: 67%), high median detectivities (MW: 4:8  1011 cm Hz1=2 Wÿ1 ; LW: 7:1  1010 cm Hz1=2 Wÿ1 ), and low median NEDTs (MW: 20 mK; LW: 7.5 mK for TSCENE ˆ 295 K and f=2.9). 6. Conclusion To summarise, despite serious competition from alternative technologies and slower progress than expected, HgCdTe is unlikely to be seriously challenged for high-performance applications, applications requiring multispectral capability and fast response. The recent successes of competing cryogenically cooled detectors are due to technological, not fundamental issues. There are good reasons to think that the steady progress in epitaxial technology would make HgCdTe devices much more a€ordable in the near future. The much higher operation temperature of HgCdTe compared to Schottky barrier devices and low-dimensional solid devices may become a decisive

argument in this case. The development of sophisticated controllable vapour phase epitaxial growth methods, such as MBE and MOCVD, has allowed the fabrication of almost ideally designed heterojunction photodiodes. Such devices can be obtained using three-dimensional gap and doping engineering, with the narrow gap absorber buried in a wide gap semiconductor. Whether the SL IR photodetectors can outperform the ``bulk'' narrow gap HgCdTe detectors is one of the most important questions for the future of IR photodetectors. In this respect, the InAs/GaInSb type II SL is the most likely candidate to replace the conventional narrow gap semiconductors. Preliminary InAs/GaInSb SLS photodiodes have led to the demonstration of prototype detectors that exhibit promising performance. However, signi®cant obstacles in material growth and device fabrication need to be addressed before its full potential can be realised. It seems to be doubtful that InAs/GaInSb SLS photodiodes will replace HgCdTe photodiodes in the MWIR and LWIR spectral ranges. However, the SLS photodiodes can ®nd application in the VLWIR spectral range operating at elevated temperature compared to extrinsic Si detectors. The main challenges facing dual-band devices are more complicated device structures, thicker and multilayer material growth, and more dicult fabrication, especially when the array size gets larger and pixel size gets smaller. In the case of HgCdTe devices, the multiple p±n junctions present more diculties in material growth, device fabrication and passivation. Several companies plan to advance simultaneous dual-band FPA technology to larger array sizes, smaller pixels, and higher performance, especially that required for low background applications. For strategic applications, special variants of dual-band input circuits are designed for improved performance at lower background photon ¯uxes. The commercial market will probably be dominated by uncooled IR FPAs, except for medical applications where high resolution and accuracy are needed. However, uncooled detectors developed so far are less sensitive than the cooled de-

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

tectors. It also has no intrinsic multicolour capability. Acknowledgements This work was partially supported by the KBN (Poland) under grant number PBZ 28.11/P6. References [1] A. Rogalski, Infrared photovoltaic detectors, Opto-Electr. Rev. 5 (1997) 205±216. [2] A. Rogalski, M. Razeghi, Narrow gap semiconductor photodiodes, Proc. SPIE 3287 (1998) 2±13. [3] D.A. Scribner, M.R. Kruer, J.M. Killiany, Infrared focal plane array technology, Proc. IEEE 79 (1991) 66±85. [4] J. Piotrowski, A. Rogalski, New generation of infrared photodetectors, Sensors and Actuat. A67 (1998) 146±152. [5] A. Rose, Concepts in Photoconductivity and Allied Problems, Interscience, New York, 1963. [6] J. Piotrowski, W. Gawron, Ultimate performance of infrared photodetectors and ®gure of merit of detector material, Infrared Phys. Tech. 38 (1997) 63±68. [7] J. Piotrowski, Hg1ÿx Cdx Te detectors, In: A. Rogalski (Ed.), Infrared Photon Detectors, SPIE Optical Engineering Press, Bellingham, pp. 391±493. [8] A. Rogalski, R. Ciupa, Long wavelength HgCdTe photodiodes: n‡ -on-p versus p-on-n structures, J. Appl. Phys. 77 (1995) 3505±3512. [9] A. Rogalski, J. Piotrowski, Intrinsic infrared detectors, Prog. Quant. Electr. 12 (1988) 87±289. [10] J. Piotrowski, W. Galus, M. Grudzie n, Near roomtemperature IR photo-detectors, Infrared Phys. 31 (1991) 1±48. [11] C.T. Elliott, N.T. Gordon, Infrared detectors, in: C. Hilsum (Ed.), Handbook on Semiconductors, vol. 4, North-Holland, Amsterdam, 1993, pp. 841±936. [12] T. Elliott, New infrared and other applications of narrow gap semiconductor, Proc. SPIE 3436 (1997) 763±775. [13] K. Adamiec, A. Rogalski, J. Rutkowski, Progress in infrared detector technology, J. Tech. Phys. 38 (1997) 431±488. [14] A. Rogalski (Ed.), Infrared Photon Detectors, SPIE Optical Engineering Press, Bellingham, 1995. [15] A. Rogalski, Infrared Detectors, Gordon and Breach, London, 2000. [16] A. Rogalski, New Ternary Alloy Systems for Infrared Detectors, SPIE Optical Engineering Press, Bellingham, 1994. [17] J.L. Lee, J.D. Kim, M. Razeghi, Novel InTlSb alloy for uncooled long-wavelength infrared photodetectors, OptoElectr. Rev. 7 (1999) 19±28.

235

[18] J.J. Lee, M. Razeghi, Exploration of InSbBi for uncooled long-wavelength infrared photodetectors, Opto-Electr. Rev. 6 (1998) 25±36. [19] M.B. Reine, K.R. Maschho€, S.P. Tobin, P.W. Norton, J.A. Mroczkowski, E.E. Krueger, The impact of characterization techniques on HgCdTe infrared detector technology, Semicon. Sci. Tech. 8 (1998) 788±804. [20] P.R. Bratt, HgCdTe heterojunctions, J. Vac. Sci. Tech. A1 (1983) 1687±1691. [21] P.R. Bratt, T.N. Casselman, Potential barriers in HgCdTe heterojunctions, J. Vac. Sci. Tech. A3 (1985) 238±245. [22] K. Zanio, K. Kay, Modeling of HgCdTe heterojunction devices, Mat. Res. Soc. Proc. 90 (1987) 39±46. [23] F.L. Madarasz, Barrier formation in graded Hg1±x Cdx Te heterojunctions, J. Appl. Phys. 62 (1987) 3267±3277. [24] F.L. Madarasz, F. Szmulowicz, Valence-band barrier formation in graded Hg1± x Cdx Te heterojunctions with a valence-band o€set induced, J. Appl. Phys. 66 (1989) 3082± 3087. [25] K. Kosai, W.A. Radford, In¯uence of barriers on charge transport across HgCdTe heterojunctions, J. Vac. Sci. Tech. A8 (1990) 1254±1259. [26] M.H. Weiler, M.B. Reine, E€ect of a valence-band barrier on the quantum eciency and background-limited dynamic resistance of compositionally geaded HgCdTe P-on-n heterojunction photodiodes, J. Electron. Mater. 24 (1995) 1329±1339. [27] S.P. Tobin, M.H. Weiler, M.A. Hutchins, T. Parodos, P.W. Norton, Advances in composition control for 16 lm LPE p-on-n HgCdTe heterojunction photodiodes for remote sensing applications at 60 K, J. Electron. Mater. 28 (1999) 596±602. [28] J.M. Arias, S.H. Shin, J.G. Pasko, R.E. DeWames, E.R. Gertner, Long and middle wavelength infrared photodiodes fabricated with Hg1ÿx Cdx Te grown by molecular beam epitaxy, J. Appl. Phys. 65 (1989) 1747±1753. [29] L.O. Bubulac, D.D. Edwall, D. McConnel, R.E. DeWames, E.R. Blazejewski, E.R. Gertner, P-on-n activated junction in LWIR HgCdTe/GaAs, Semicond. Sci. Tech. 5 (1990) S45±S48. [30] J.M. Arias, M. Zandian, J.G. Pasko, S.H. Shin, L.O. Bubulac, R.E. DeWames, W.E. Tennant, Molecular-beam epitaxy growth and in situ arsenic doping of p-on-n HgCdTe heterojunctions, J. Appl. Phys. 69 (1991) 2143± 2148. [31] G.N. Pultz, P.W. Norton, E.E. Krueger, M.B. Reine, Growth and characterization of p-on-n HgCdTe liquidphase epitaxy heterojunction material for 11±18 lm applications, J. Vac. Sci. Tech. B9 (1991) 1724±1730. [32] C.C. Wang, Mercury cadmium telluride junctions grown by liquid phase epitaxy, J. Vac. Sci. Tech. B9 (1991) 1740± 1745. [33] M.B. Reine, K.R. Maschho€, S.P. Tobin, P.W. Norton, J.A. Mroczkowski, E.E. Krueger, The impact of characterization techniques on HgCdTe infrared detector technology, Semicon. Sci. Tech. 8 (1993) 788±804.

236

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

[34] P.R. Norton, Infrared image sensors, Opt. Engng. 30 (1991) 1649±1663. [35] R.E. DeWames, J.M. Arias, L.J. Kozlowski, G.M. Williams, An assessment of HgCdTe and GaAs/GaAlAs technologies for LWIR infrared imagers, Proc. SPIE 1735 (1992) 2±16. [36] J.M. Arias, J.G. Pasko, M. Zandian, L.J. Kozlowski, R.E. DeWames, Molecular beam epitaxy HgCdTe infrared photovoltaic detectors, Opt. Engng. 33 (1994) 1422±1428. [37] M.B. Reine, E.E. Krueger, P. OÕDette, C.L. Terzis, B. Denley, J. Hartley, J. Rutter, D.E. Kleinmann, Advances in 15 lm HgCdTe photovoltaic and photoconductive detector technology for remote sensing, Proc. SPIE 2816 (1996) 120±137. [38] Y. Nemirovsky, N. Amir, Sufraces/interfaces of narrowgap II±VI compounds, in: P. Capper (Ed.), Chapman @ Hall, London, 1997, pp. 291±326. [39] W.E. Tennant, C.A. Cockrum, J.B. Gilpin, M.A. Kinch, M.B. Reine, R.P. Ruth, Key issues in HgCdTe-based focal plane arrays: a industry perspective, J. Vac. Sci. Tech. B10 (1992) 1359±1369. [40] T. Tung, L.V. DeArmond, R.F. Herald, P.E. Herning, M.H. Kalisher, D.A. Olson, R.F. Risser, A.P. Stevens, S.J. Tighe, State of the art of Hg-melt LPE HgCdTe at Santa Barbara Research Center, Proc. SPIE 1735 (1992) 109±134. [41] S.M. Johnson, T.J. deLyon, C.A. Cockrum, W.J. Hamilton, T. Tung, F.I. Gesswein, B.A. Baumgratz, L.M. Ruzicka, O.K. Wu, J.A. Torh, Direct growth of CdZnTe/ Si substrates for large-area HgCdTe infrared focal plane arrays, J. Electr. Mat. 24 (1995) 467±473. [42] T.J. de Lyon, S.M. Johnson, C.A. Cockrum, O.K. Wu, W.J. Hamilton, G.S. Kamath, CdZnTe on Si (0 0 1) and Si (1 1 2): Direct MBE growth for large-area HgCdTe infrared focal-plane array applications, J. Electrochem. Soc. 141 (1994) 2888±2893. [43] T.J. de Lyon, R.D. Rajavel, J.E. Jensen, O.K. Wu, S.M. Johnson, C.A. Cockrum, G.M. Venzor, Heteroepitaxy of HgCdTe(1 1 2) infrared detector structures on Si(1 1 2) substrates by molecular-beam epitaxy, J. Electron. Mater. 25 (1996) 1341±1346. [44] T.J. de Lyon, R.D. Rajavel, J.E. Jensen, O.K. Wu, J.A. Vigil, S.M. Johnson, C.A. Cockrum, G.M. Venzor, Growth of HgCdTe IR detector structures on Si substrates, Proc. SPIE 2816 (1996) 29±41. [45] P.S. Wijewarnasuriya, M. Zandian, D.D. Edwall, M.V. McLevige, C.A. Chen, J.G. Pasko, H. Hildebrandt, A.C. Chen, J.M. Arias, A.I. DÕSouza, S. Rujirawat, S. Sivananthan, MBE p-on-n Hg1± x Cdx Te heterostructure detectors on silicon substrates, J. Electron. Mater. 27 (1998) 546± 549. [46] T.J. de Lyon, R.D. Rajavel, J.A. Vigil, J.E. Jensen, O.K. Wu, C.A. Cockrum, S.M. Johnson, G.M. Venzor, S.L. Bailey, I. Kasai, W.L. Ahlgren, M.S. Smith, Molecularbeam epitaxial growth of HgCdTe infrared focal-plane arrays on silicon substrates for midwave infrared applications, J. Electron. Mater. 27 (1998) 550±555.

[47] M.C. Chen, R.S. List, D. Chandra, M.J. Bevan, L. Colombo, H.F. Schaake, Key performance-limiting defects in p-on-n HgCdTe heterojunction infrared photodiodes, J. Electron. Mater. 25 (1996) 1375±1382. [48] R.S. List, J.H. Tregilgas, A.M. Turner, J.D. Beck, J.C. Ehmke, Electrical e€ects of subgrain boundaries twins dislocations and Te precipitation on LWIR HgCdTe photodiode arrays, Proc. SPIE 2228 (1994) 274±282. [49] R.S. List, Formation and electrical e€ects of process induced dislocations in HgCdTe, J. Vac. Sci. Tech. B10 (1992) 1651±1657. [50] D.L. Smith, C. Mailhiot, Proposal for strained type II superlattice infrared detectors, J. Appl. Phys. 62 (1987) 2545±2548. [51] C. Mailhiot, Far-infrared materials based on InAs/GaInSb type II, strained-layer superlattices, in: M.O. Manasreh (Ed.), Semiconductor Quantum Wells and Superlattices for Long-Wavelength Infrared Detectors, Artech House, Boston, 1993, pp. 109±138. [52] C.H. Grein, P.M. Young, M.E. Flatte, H. Ehrenreich, Long wavelength InAs/InGaSb infrared detectors: Optimization of carrier lifetimes, J. Appl. Phys. 78 (1995) 7143± 7152. [53] H. Ehrenreich, C.H. Grein, R.H. Miles, M.E. Flatte, Reply to ``Comment on ``Temperature limits on infrared detectivities of InAs/Inx Ga1± x Sb superlattices and bulk Hgx Cd1± x Te, [J. Appl. Phys. 80 (1996) 2542]'' '', J. Appl. Phys. 80 (1996) 2545±2546. [54] A. Rogalski, Comparison of the performance of quantum well and conventional bulk infrared photodetectors, Infrared Phys. Tech. 38 (1997) 295±310. [55] J. Piotrowski, A. Rogalski, Comment on ``Temperature limits on infrared detectivities of InAs/Inx Ga1±x Sb superlattices and bulk Hgx Cd1±x Te [J. Appl. Phys. 74 (1993) 4774]'', J. Appl. Phys. 80 (1996) 2542±2544. [56] E.R. Youngdale, J.R. Meyer, C.A. Ho€man, F.J. Bartoli, C.H. Grein, P.M. Young, H. Ehrenreich, R.H. Miles, D.H. Chow, Auger lifetime enhancement in InAs±Ga1±x Inx Sb superlattices, Appl. Phys. Lett. 64 (1994) 3160±3162. [57] J.L. Johnson, The InAs/GaInSb strained layer superlattice as an infrared detector material: an overview, Proc. SPIE, submitted for publication. [58] J.L. Johnson, L.A. Samoska, A.C. Gossard, J.L. Merz, M.D. Jack, G.H. Chapman, B.A. Baumgratz, K. Kosai, S.M. Johnson, Electrical and optical properties of infrared photodiodes using the InAs/Ga1± x Inx Sb superlattice in heterojuntions with GaSb, J. Appl. Phys. 80 (1996) 1116± 1127. [59] F. Fuchs, U. Weimer, E. Ahlswede, W. Pletschen, J. Schmitz, M. Walther, InAs/Ga1± x Inx Sb infrared superlattice photodiodes for infrared detection, Proc. SPIE 3287 (1998) 14±21. [60] M.H. Young, D.H. Chow, A.T. Hunter, R.H. Miles, Appl. Surface Sci. 123/124 (1998) 395±399. [61] F. Fuchs, W. Pletschen, J. Schmitz, M. Walther, H. Gullich, N. Herres, S. Muller, InAs/GaInSb infrared

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

[62] [63]

[64] [65]

[66]

[67]

[68] [69] [70] [71] [72]

[73] [74] [75]

[76]

superlattice diodes: correlation between surface morphology and electrical performance, Proc SPIE 3794 (1999). K. Vural, Mercury cadmium telluride short- and mediumwavelength infrared staring focal plane arrays, Opt. Engng. 26 (1987) 201±208. L.J. Kozlowski, R.B. Bailey, S.A. Cabelli, D.E. Cooper, I.S. Gergis, A.C. Chen, W.V. McLevige, G.L. Bostrup, K. Vural, W.E. Tennant, P.H. Howard, High-performance 5 lm 640  480 HgCdTe-on-sapphire focal plane arrays, Opt. Engng. 33 (1994) 54±63. L.J. Kozlowski, HgCdTe focal plane arrays for high performance infrared cameras, Proc. SPIE 3179 (1997) 200±211. L.J. Kozlowski, K. Vural, D.Q. Bui, R.B. Bailey, D.E. Cooper, D.M. Stephenson, Status and direction of PACE1 HgCdTe FPAs for astronomy, Proc. SPIE 1946 (1993) 148±160. L.J. Kozlowski, K. Vural, S.C. Cabelli, C.Y. Chen, D.E. Cooper, G.L. Bostrup, D.M. Stephenson, W.L. McLevige, R.B. Bailey, K. Hodapp, D. Hall, W.E. Kleinhans, 2.5 lm PACE-1 HgCdTe 1024  1024 FPA for infrared astronomy, Proc. SPIE 2268 (1994) 353±364. L.O. Bubulac, W.E. Tennant, J.G. Pasko, L.J. Kozlowski, M. Zandian, M.E. Motamedi, R.E. DeWames, J. Bajaj, N. Nayar, W.V. McLevige, N.S. Gluck, R. Melendes, D.E. Cooper, D.D. Edwall, J.M. Arias, R. Hall, High performance SWIR HgCdTe detector arrays, J. Electron. Mater. 26 (1997) 649±655. A. Rogalski, R. Ciupa, Performance limitation of short wavelength infrared InGaAs and HgCdTe photodiodes, J. Electron. Mater. 28 (1999) 630±636. R.G. Humpreys, Radiative lifetime in semiconductors for infrared detection, Infrared Phys. 23 (1983) 171±175. R.G. Humpreys, Radiative lifetime in semiconductors for infrared detection, Infrared Phys. 26 (1986) 337±342. C.H. Grein, H. Ehrenreich, E. Runge, Radiative lifetime in semiconductors: in¯uence of photon recycling, Proc. SPIE 2999 (1998) 11±18. L.J. Kozlowski, K. Vural, J.M. Arias, W.E. Tennant, R.E. DeWames, Performance of HgCdTe, InGaAs and quantum well GaAs/AlGaAs staring infrared focal plane arrays, Proc. SPIE 3182 (1997) 2±13. G.H. Olsen, M.J. Cohen, Applications of near-infrared imaging, Proc. SPIE 3379 (1998) 300±306. A. Krier, Y. Mao, High performance uncooled InAsSbP/ InGaAs photodiodes for the 1.8±3.4 lm wavelength range, Infrared Phys. Tech. 38 (1997) 397±403. R.M. Lin, S.F. Tang, S.C. Lee, C.H. Kuan, G.S. Chen, T.P. Sun, J.C. Wu, Room temperature unpassivated InAs p±i±n photodetectors grown by molecular beam epitaxy, IEEE Trans. Electron Dev. 44 (1997) 209± 213. R.E. DeWames, D.D. Edwall, M. Zandian, L.O. Bubulac, J.G. Pasko, W.E. Tennant, J.M. Arias, A. DÕSouza, Dark current generating mechanisms in short wavelength infrared photovoltaic detectors, J. Electron. Mater. 27 (1998) 722±726.

237

[77] P.R. Norton, Status of infrared detectors, Proc. SPIE 3379 (1998) 102±114. [78] P.R. Norton, Infrared detectors in the next millennium, Proc. SPIE 3698 (1999) 652±665. [79] M.Z. Tidrow, QWIP and MCT for long wavelength and multicolor focal plane array applications, presented during Sixth Annual AIAA/BMDO Technology Readiness Conference and Exhibit, San Diego, CA, 1997. [80] A. Rogalski, Assessment of HgCdTe photodiodes and quantum well infrared photoconductors for long wavelength focal plane arrays, Infrared Phys. Tech. 40 (1999) 279±294. [81] R.D. Rajavel, D.M. Jamba, O.K. Wu, J.E. Jensen, J.A. Wilson, E.A. Patten, K. Kosai, P. Goetz, G.R. Chapman, W.A. Radford, High performance HgCdTe two-color infrared detectors grown by molecular beam epitaxy, J. Crystal Growth 175 (1997) 653±658. [82] R.D. Rajavel, D.M. Jamba, J.E. Jensen, O.K. Wu, P.D. Brewer, J.A. Wilson, J.L. Johnson, E.A. Patten, K. Kosai, J.T. Caul®eld, P.M. Goetz, Molecular beam epitaxial growth and performance of integrated multispectral HgCdTe photodiodes for the detection of mid-wave infrared radiation, J. Crystal Growth 184 (1998) 1272± 1278. [83] R.D. Rajavel, D.M. Jamba, J.E. Jensen, O.K. Wu, J.A. Wilson, J.L. Johnson, E.A. Patten, K. Kasai, P.M. Goetz, S.M. Johnson, Molecular beam epitaxial growth and performance of HgCdTe-based simultaneous-mode two-color detectors, J. Electron. Mater. 27 (1998) 747± 751. [84] J.L. Johnson, Two-color HgCdTe devices for multispectral applications, Extended Abstracts of The 1999 US Workshop on the Physics and Chemistry of II±VI Materials, Las Vegas, 1999, pp. 101±104. [85] M.B. Reine, P.W. Norton, R. Starr, M.H. Weiler, M. Kestigian, B.L. Musicant, P. Mitra, T. Schimert, F.C. Case, I.B. Bhat, H. Ehsani, V. Rao, Independentlyaccessed back-to-back HgCdTe photodiodes: A new dual-band infrared detector, J. Electron. Mater. 24 (1995) 669±679. [86] P. Mitra, S.L. Barnes, F.C. Case, M.B. Reine, P. OÕDette, R. Starr, A. Hairston, K. Kuhler, M.H. Weiler, B.L. Musicant, MOCVD of band gap-engineered HgCdTe p±n±N±P dual-band infrared detector arrays, J. Electron. Mater. 26 (1997) 482±487. [87] M.B. Reine, A. Hairston, P. OÕDette, S.P. Tobin, F.T.J. Smith, B.L. Musicant, P. Mitra, F.C. Case, Simultaneous MW/LW dual-band MOCVD HgCdTe 64  64 FPAs, Proc. SPIE 3379 (1998) 200±212. [88] H.K. Pollehn, J. Ahearn, Multi-domain smart sensors, Proc. SPIE 3698 (1999) 420±426. [89] J.C. Campbell, A.G. Dentai, T.P. Lee, C.A. Burrus, Improved two-wavelength demultiplexing InGaAsP photodetector, IEEE J. Quantum Electron. QE-16 (1980) 601±603. [90] J.A. Wilson, E.A. Patten, G.R. Chapman, K. Kosai, B. Baumgratz, P. Goetz, S. Tighe, R. Risser, R. Herald, W.A.

238

A. Rogalski / Infrared Physics & Technology 41 (2000) 213±238

Radford, T. Tung, W.A. Terre, Integrated two-color detection for advanced FPA applications, Proc. SPIE 2274 (1994) 117±125. [91] S.M. Johnson, D.R. Rhiger, J.P. Rosbeck, J.M. Peterson, S.M. Taylor, M.E. Boyd, E€ect of dislocations on the electrical and optical properties of long-wavelength infra-

red HgCdTe photovoltaic detectors, J. Vac. Sci. Tech. B10 (1992) 1499±1506. [92] K. J ozwikowski, A. Rogalski, E€ect of dislocations on performance of LWIR HgCdTe photodiodes, Extended Abstracts of The 1999 US Workshop on the Physics and Chemistry of II±VI Materials, Las Vegas, 1999, pp. 95±98.