Outlook on Quantum Dot Infrared Photodetectors
ISSN 1060-992X, Optical Memory and Neural Networks (Information Optics), 2009, Vol. 18, No. 3, pp. 234–252. ? Allerton Press, Inc., 2009.
234
Outlook on Quantum Dot Infrared Photodetectors 1
A. Rogalski
Institute of Applied Physics, Military University of Technology, 2 Kaliskiego Str., 00-908 Warsaw, Poland
Received September 23, 2008; in ?nal form, June 8, 2009
Abstract —This paper reviews the present status and gives outlook on future developments of quantum dot infrared photodetectors (QDIPs). At the beginning the paper summarizes the fundamental properties of QDIPs. Next, an emphasize is put on their potential developments. Investigations of the performance of QDIPs as compared to other types of infrared photodetectors are presented. A model is based on fun-damental performance limitations enabling a direct comparison between different infrared material tech-nologies. It is assumed that the performance is due to thermal generation in the active detector’s region.In comparative studies, the HgCdTe photodiodes, quantum well infrared photodetectors (QWIPs), type II superllatice photodiodes, Schottky barrier photoemissive detectors, doped silicon detectors, and high temperature superconductor detectors are considered.
Theoretical predictions indicate that only type II superlattice photodiodes and QDIPs are expected to compete with HgCdTe photodiodes. QDIPs theoretically have several advantages compared with QWIPs including the normal incidence response, lower dark current, higher operating temperature,higher responsivity and detectivity. The operating temperature for HgCdTe detectors is higher than for other types of photon detectors.
Comparison of QDIP performance with HgCdTe detectors gives evidence that the QDIP is suitable for high operation temperature. It can be expected that improvement in technology and design of QDIP detectors will make it possible to achieve both high sensitivity and fast response useful for practical application at room temperature.
Finally, issue associated with the development and exploitation of multispectral photodetectors from this new material is discussed. Discussions is focused on most recently on-going detector technology efforts in fabrication both photodetectors and focal plane arrays.
Key words : quantum dot infrared photodetectors, quantum well infrared photodetectors, HgCdTe pho-todiodes, type-II seperlattices, detectivity, focal plane arrays, multispectral detectors. DOI: 10.3103/S1060992X0903014X
1. INTRODUCTION
Since the initial proposal by Esaki and Tsu in 1970 [1] and the advent of molecular beam epitaxy (MBE), the interest in semiconductor low-dimensional solids has increased continuously over the years, driven by technolog-ical challenges, new physical concepts and phenomena as well as promising applications. A new class of materials with unique optoelectronic properties has been developed. Zero-dimensional quantum con?ned semiconductor heterostructures have been investigated theoretically and experimentally for some time [2–4]. At present, nearly defect-free quantum dot devices can be fabricated reliably and reproducibly. Also new types of infrared photode-tectors taking advantage of the quantum con?nement obtained in semiconductor heterostructures have been emerged.
As it was indicated by Kinch [5], photon detectors can be divided into two broad classes, namely majority and minority carrier devices. We can distinguish six infrared (IR) material systems:(1) Direct bandgap semiconductors—minority carriers.Binary alloys: InSb, InAs
? Ternary alloys: HgCdTe, InGaAs
? Type II, III superlattices: InAs/GaInSb, HgTe/CdTe
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
235
? Si:As, Si:Ga, Si:Sb ? Ge:Hg, Ge:Ga
(3) Type I superlattices—majority carriers ? GaAs/AlGaAs QWIPs
(4) Silicon Schottky barriers—majority carriers ? PtSi, IrSi
(5) Quantum dots—majority carriers ? InAs/GaAs QDIPs
(6) High temperature superconductors—minority carriers
All of these material systems have been seriously players in the infrared marketplace with the exception of the high temperature superconductors and quantum dot infrared photodetectors (QDIPs). The dates given in Fig. 1show the chronology of signi?cant development efforts on the materials mentioned. First observations of intersub-level transitions in the far infrared were reported in the early 1990s, either in InSb-based electrostatically de?ned quantum dots [6] or in structured two-dimensional (2-D) electron gas [7]. The ?rst QDIP was demonstrated in 1998[8]. Ever since great progress has been made in their development and performance characteristics [9, 10] and in their applications to thermal imaging focal plane arrays [11].
The beginning of the interest in quantum dot research can be traced back to a suggestion by Arakawa and Sakaki in 1982 [2] that the performance of semiconductor lasers could be improved by reducing the dimensionality of the active regions of these devices. Initial efforts at reducing the dimensionality of the active regions focused on using ultra?ne lithography coupled with wet or dry chemical etching to form 3-D structures. It was soon realized, how-ever, that this approach introduced defects (high density of surface states) that greatly limited the performance of such quantum dots. Initial efforts were mainly focused on the growth of InGaAs nanometer-sized islands on GaAs substrates. In 1993, the ?rst epitaxial growth of defect-free quantum-dot nanostructures was achieved by using MBE [12]. Most of the practical quantum-dot structures today are synthesized both by MBE and MOCVD.
Under certain growth conditions, when the thickness of the ?lm with the larger lattice constant exceeds a certain critical thickness, the compressive strain within the ?lm is relieved by the formation of coherent island. These islands may be quantum dots. Coherent quantum-dot islands are generally formed only when the growth proceeds in what is known as Stranski-Krastanow growth model [13]. The onset of the transformation of the growth process
T h e r m a l d e t e c t o r s
Detector roadmap
Video
w o -c o l o u r F P A s
o l o m e t e r F P A s S i :X
V e r y l a r g e F P A s
MWIR contact
LWIR contact
SWIR cap layer Substrate
MWIR
LWIR
1 st Gen, Scan to image
2 nd Gen
FPA+ROIC
3 rd Gen FPA+ROIC
T l 2 S
P b S P b S e
e :X n S b
g C d T e
P b S n T e S i :X /C C D P l S i /C C D
H g C d T e /C C D
g C d T e S P R I T E
n G a A s
W I P
y r o e l e c t r i c F P A s
D I P
E M S
F P A s o u r -c o l o u r F P A s
w o -c o l o u r Q D I P F P A s
194019501960
1970198019902000
2010
Fig. 1. History of the development of infrared detectors and systems. Three generation systems can be considered for prin-cipal military and civilian applications: 1 st Gen (scanning systems), 2 nd Gen (staring systems—electronically scanned) and rd
Gen (multicolour functionality and other on-chip functions).
236
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
from a 2-D layer-by-layer growth mode to a 3-D island growth mode results in a spotty RHEED pattern. This is,in contrast to the conventional streaky pattern, generally observed for the layer-by-layer growth mode. The transi-tion typically occurs after the deposition of a certain number of monolayers. For InAs on GaAs, this transition occurs after about 1.7 monolayers of InAs have been grown; this is the onset of islanding and, hence, quantum-dot formation.
The most advanced III-V infrared detectors, which utilize intersubband or subband to continuum transitions in quantum wells, are GaAs/AlGaAs quantum well infrared photodetectors (QWIPs). The imaging performance of focal plane arrays fabricated with this material system is comparable to the state of art of HgCdTe [14, 15].
This paper summarizes the fundamental properties of QDIPs. The intent is to concentrate on device approaches and present stage of development. A secondary aim is to compare the potential QDIP performance with different material systems used in infrared detector technology. Our intention is to concentrate on fundamental phenomena and minimize any confusion that might exist within the minds of scientists. The paper completes two previously published papers by Kinch [5] and Phillips [16].
2. ANTICIPATED ADV ANTAGES OF QDIPS
The success of quantum well structures for infrared detection applications has stimulated the development of QDIPs. In general, QDIPs are similar to QWIPs but with the quantum wells replaced by quantum dots, which have size con?nement in all spatial directions.
Figure 2 shows the schematic layers of a QWIP and a QDIP . In both cases, the detection mechanism is based on the intraband photoexcitation of electrons from con?ned states in the conduction band wells or dots into the con-tinuum. The emitted electrons drift towards the collector in the electric ?eld provided by the applied bias, and pho-tocurrent is created. It is assumed, that the potential pro?le at the conduction band edge along the growth direction for both structure have a similar shape as shown in Fig. 2b. In practice, since the dots are spontaneously self-assem-bled during growth, they are not correlated between multilayers in active region.
Two types of QDIP structures have been proposed: conventional structure (vertical) and lateral structure (Fig. 3). In a vertical QDIP , the photocurrent is collected through the vertical transport of carriers between top and bottom contacts (see Fig. 2a). The device heterostructure comprises repeated InAs QD layers buried between GaAs barriers with top and bottom contact layers at active region boundaries. The mesa height can vary from 1 to 4 μ m depending on the device heterostructure. The quantum dots are directly doped (usually with silicon) in order to pro-vide free carriers during photoexcitation, and an AlGaAs barrier can be included in the vertical device heterostruc-ture in order to block dark current created by thermionic emission [18, 19].
In addition to the standard InAs/GaAs QDIP , several other heterostructure designs have been investigated for use as IR photodetectors [9, 10]. An example is InAs QDs embedded in a strain-relieving InGaAs quantum well which are known as dot-in-a-wall (DWELL) heterostructures (see Fig. 4) [11, 20]. This device offers two advan-tages: challenges in wavelength tuning through dot-size control can be compensated in part by engineering the quantum wel sizes, which can be controlled precisely and quantum wells can trap electrons and aid in carrier cap-ture by QDs, thereby facilitating ground state re?lling. Figure 4b shows DWELL spectral tuning by varying well geometry.
The lateral QDIP (see Fig. 3b) collects photocurrent through transport of carriers across a high-mobility channel between two top contacts, operating much like a ?eld-effect transistor. As previously, again AlGaAs barriers are
(a)
(b)
Growth Top contact
Barrier Well Bottom contact
Dots Emitter
Injection from Trapping
Emission
Photocurrent
Collector
Under bias direction
Barrier Barrier Barrier
contact
QWIP
QDIP
xN
h ν
xN
Fig. 2.
Schematic layers of QWIP and QDIP (a) and potential pro?le for both structures under bias (b). For QDIP, in?uence of wetting layer is neglected (after [17]).
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
237
present, but instead of blocking the dark current, these barriers are used to both modulation-dope the quantum dots and to provide the high-mobility channel. Lateral QDIPs have demonstrated lower dark currents and higher oper-ating temperatures than vertical QDIPs since the major components of the dark current arise from interdot tunnel-ling and hopping conduction [22]. However, these devices will be dif?cult to incorporate into a FPA hybrid-bump bonded to a silicon read-out circuit. Because of this, more efforts is directed to improve the performance of vertical QDIPs, which are more compatible with commercially available read-out circuits.
The quantum-mechanical nature of QDIPs leads to several advantages over QWIPs and other types of IR detec-tors that are available. As in the HgCdTe, QWIP and type II superllatice technologies, QDIPS provide multi-wave-length detection. However, QDs provide many additional parameters for tuning the energy spacing between energy levels, such as QD size and shape, strain, and material composition.
The potential advantages in using QDIPs over quantum wells are as follows:
? Intersubband absorption may be allowed at normal incidence (for n -type material). In QWIPs, only transitions polarized perpendicularly to the growth direction are allowed, due to absorption selection rules. The selection rules in QDIPs are inherently different, and normal incidence absorption is observed.
? Thermal generation of electrons is signi?cantly reduced due to the energy quantization in all three dimensions.As a result, the electron relaxation time from excited states increases due to phonon bottleneck. Generation by LO phonons is prohibited unless the gap between the discrete energy levels equals exactly to that of the phonon. This prohibition does not apply to quantum wells, since the levels are quantized only in the growth direction and a con-tinuum exists in the other two directions (hence generation-recombination by LO phonons with capture time of few picoseconds). Thus, it is expected that S/N ratio in QDIPs will be signi?cantly larger than that of QWIPs.
? Lower dark current of QDIPs is expected than of HgCdTe detectors and QWIPs due to 3-D quantum con?ne-ment of the electron wavefunction.
Both the increased electron lifetime and the reduced dark current indicate that QDIPs should be able to provide high temperature operation. In practice, however, it has been a challenge to meet all of above expectations.
IR radiation
Ohmic contact Active region Quantum dot layer Wetting layer Contact layer
Buffer layer Substrate
AlGaAs δ -doped to n = 10 10
cm 4
Repeated
(a)
(b)
several times
(Multi periods)
Ohmic contact
Ohmic contact Ohmic contact
Direction of photoelectron
current flow
Direction of photoelectron current flow IR radiation
Ohmic contact Barrier layer Spacer layer
GaAs n = 4 × 10 10 cm 4
GaAs spacer GaAs barrier InGaAs/GaAs wetting layer
S.I. GaAs substrate
InAs QDs
Fig. 3. Schematic diagram of conventional (a) and lateral (b) quantum dot detector structures.
GaAs InGaAs N o r m a l i z e d r e s p o n s i v i t y
Wavelength, μ m
(a)
(b)
IR
InAs QD
GaAs barrier
barrier
well
1.21.00.80.60.40.203
5
7
9
1113 Fig. 4.
DWELL infrared detector: (a) the operation mechanism, (b) experimentally measured spectral tunability by varying well width from 55 to 100 ? (after [21]).
238
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
Carrier relaxation times in QDs are longer than the typical 1–10 ps measured for quantum wells. It is predicted that the carrier relaxation time in QDs is limited by electron-hole scattering [23], rather than phonon scattering. For QDIPs, the lifetime is expected to be even larger, greater than 1 ns, since the QDIPs are majority carrier devices due to absence of holes.
The main disadvantage of the QDIP is the large inhomogeneous linewidth of the quantum-dot ensemble vari-ation of dot size in the Stranski-Krastanow growth mode [16, 24]. As a result, the absorption coef?cient is reduced,since it is inversely proportional to the ensemble linewidth. Large, inhomogeneously broadened linewidth has a deleterious effect on QDIP performance. Subsequently, the quantum ef?ciency QD devices tend to be lower than what is predicted theoretically. V ertical coupling of quantum-dot layers also reduces the inhomogeneous linewidth of the quantum-dot ensemble; however, it may also increase the dark current of the device, since carriers can tunnel through adjacent dot layers more easily. As in other type of detectors, also nonuniform dopant incorporation adversely affects the performance of the QDIP . Therefore, improving QD uniformity is a key issue in the increasing absorption coef?cient and improving the performance. Thus, the growth and design of unique QD heterostructure is one of the most important issues related to achievement of state-of-the art QDIP performance.3. PERFORMANCE LIMITS OF INFRARED PHOTODETECTORS
The total generation rate of infrared detector is a sum of the optical and thermal generation
G = G th + G op .
(1)
The optical generation may be due to the signal or background radiation. For IR detectors, usually background radiation is higher compared to the signal radiation. If the thermal generation is reduced much below the back-ground level, the performance of the device is determined by the background radiation (BLIP conditions for Back-ground Limited Infrared Photodetector). This condition can be described as [5]
> n th ,(2)
where n th is the density of thermal carriers at the temperature T , τ is the carrier lifetime, B is the total back-ground photon ?ux density (unit cm –2 s –1 ) reaching the detector, and t is the detector’s thickness. Re-arrang-ing, we have for the BLIP requirements
(3)
i.e., the photon generation rate per unit volume needs to be greater than the thermal generation rate per unit
volume. The carriers can be either majority or minority in nature. Using η = α t , where α is the absorption coef?cient in the material, we obtain
(4)
The normalized thermal generation, G th = n th /( ατ ), predicts the ultimate performance of any infrared material
and can be used to compare the relative performance of different materials as a function of temperature and energy gap (cutoff wavelength).
It should be noted that the importance of the thermal generation rate as a material ?gure of merit was recognized for the ?rst time by Long [25]. It was used in many papers by English workers [26, 27] related to high operating temperature (HOT) detectors. Equation (4) was introduced by Kinch [5], which is the thermal generation rate within 1/ α depth per unit of area, as the ?gure of merit. This formula is actually the inverse α / G th ?gure of merit previously proposed by Piotrowski and Gawron [28].
In further considerations we use a simple set of fundamental detector parameters described in excellent Kinch’s paper [5] to compare the performance of different material systems used in infrared detector technology. In the case of QDIPs, a model developed by Phillips is adapted [16].
3.1. QDIP Model
Figure 5 shows a schematic view of the QDIP structure under considerations. Simple estimation indicates that the quantum dot density δ = 1/ s 2 , where s is the interdot spacing. We will consider a planar array of quantum dots with conduction band structure containing two con?ned energy levels ( E 1 and E 2
) and the excited state transition coinciding with the barrier conduction band minimum.
ηΦB τ
t
-------------ηΦB t ----------n th τ
-----.>ΦB n th ατ
------>G th .=
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
239
Due to discrete nature of QDs, the ?ll factor F should be included for optical absorption in quantum dots. This factor can be estimated in a simple way as
(5)
where V is the quantum dot volume.
For self-assembled QDs, a Gaussian distribution has been observed for the electronic and optical spectra. Phil-lips modelled the absorption spectra for an ensemble of QDs using a Gaussian line shape in the shape
(6)
where α 0 is the maximum absorption coef?cient, n 1 is the areal density of electrons in the quantum dot
ground state, δ is the quantum dot density, and E g = E 2 – E 1 is the energy of the optical transition between ground and excited states in the QDs. It should be noticed that Eq. (6) estimates the absorption coef?cient for the necessary presents of electrons in the QD ground state.
The optical absorption between the ground and excited levels is found to be have a value [29]
in cm –1 (7)
where σ is the linewidth of the transition in meV . Equation (7) indicates the tradeoff between the absorption coef?cient and the absorption linewidth, σ .
The expressions σ QD and σ ens
are the standard deviations in the Gaussian line shape for intraband absorption in a single quantum dot and for the distribution in energies for the QD ensemble, respectively. The terms n 1/δ and σQD /σens describe a decrease in absorption due to absence of available electrons in the QD ground state and inho-mogeneous broadening, respectively.
To calculate thermal distribution of carrier density, the Fermi distribution is used. Then, the electron densities in the QD sheet for the energy level n is given by
(8)
where g is the degeneracy factor for the energy level, E n is the mean energy, σ is the standard deviation in
energy for the Gaussian line shape (to describe the spread of QD energy levels, again a Gaussian distribution with standard deviation σ is used [30]). Because, however, σ < E g = E 2 – E 1, the Gaussian line shape function describing carrier densities in QD ground and excited states has little effect, and then Eq. (8) can be simpli-?ed to
(9)
F V 3s
-------,
=αE ()α0n 1δ----σQD σens ---------E E g –()2
σens
2---------------------–,exp =α0 3.5105
×σ
---------------------,
≈n n g δ
πσ----------E E n –()2
σ
2
---------------------–f E n ()
exp E ,d ∫
=
n n g δF E n .()
=(a)
(b)E c
E 2
h ν
E 1
g
S
Fig. 5. Schematic view of the quantum dot array (a) and conduction band structure of the dot (b) (after [16]).
240
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
Next, taking into account the charge neutrality condition, the 2-D carrier densities may be given by
(10)
where N d is the sheet density dopant level. We assume a degeneracy of g = 2 (two spin states) for the QD ground state, and g = 8 (fourfold degeneracy and two-spin state each [31]) for the quantum dot excited state.As it was mentioned previously, the excited state coincides with the conduction band minimum of the barrier material; E c = E 2. Then, the thermal carrier density is n 2 + n b , where n b is the carrier density in the conduction band.
3.2. Normalized Dark Current
The normalized dark current density, given by G th q , is presented as [see Eq. (4)]
(11)
In the calculation we assume the material parameters chosen by Phillips [16]: α0 = 5 × 104 cm –2, V =5.3 × 10–19 cm –3, δ = 5 × 10 cm –2, τ = 1 ns, N d = 1 × 1011 cm –2 and the detector thickness t = 1/α0. These parameters are representative for self-assembled InAs/GaAs quantum dots reported in the literature. The dopand concentration corresponds to two electrons per QD. In further analysis, for clarity, we assume that inhomogeneous broadening of the dot ensemble is neglected (σQD /σens = 1).
The normalized dark current densities for the various materials used in infrared detector technologies in LWIR spectral region (E g = 0.124 eV , λc = 10 μm) are shown in Fig. 6. In addition, the f /2 background ?ux current density is also shown. The extrinsic silicon, the high temperature superconductors (HTSC) and the photoemissive (silicon Schottky barrier) detectors are hypothetical, but are included for comparison. In the calculations, carried out for different material systems we have followed the procedures used in Kinch’s paper [5], except QDIPs where the Phillips’ model is used [16].
In the MWIR and LWIR regions, the dominant position have HgCdTe photodiodes. QWIPs are mainly used in LWIR tactical systems operating at lower temperature, typically 65–70 K, where cooling is not an issue. Large detector arrays with more than one million detector elements are fabricated by several manufacturers using these material systems. Beyond 15 μm, good performance is achieved using extrinsic silicon detectors. These detectors are termed impurity band conduction (IBC) detectors and found niche market for the astronomy and civil space communities because HgCdTe has not yet realized its potential at low temperatures and reduced background.Figure 6 displays that tuneable bandgap alloy, HgCdTe, demonstrates the highest performance (the lowest dark current/thermal generation and the highest BLIP operating temperature). These estimations are con?rmed by
N d n 1n 2n b ++2δf E 1()8δf E 2()g 2D
E ()f E c ()E ,
d 0
∞
∫
++==J dark G th q q σδt
α0n 1F τ
-----------------n 2n b +().
=
=N o r m a l i z e d d a r k c u r r e n t d e n s i t y , A /c m 2
Temperature, K
10–1
10–3
10–5
10–7
10–9
Fig. 6. Temperature dependence of the normalized dark current of various LWIR material technologies. The f /2 background ?ux current density is also shown.
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
241
experimental data [32, 33]. For very uniform QD ensembles, the QDIP performance can be close to HgCdTe one and potentially can exceeds that of HgCdTe in the region of high operation temperatures.
3.3. Detectivity
The normalized dark current, J dark = G th q , directly determines thermal detectivity
D * = (12)
Figure 7 compares the thermal detectivties of various photodetectors with cutoff wavelength in MWIR (λc = 5 μm) and LWIR (λc = 10 μm) regions. The assumed typical quantum ef?ciencies are indicated in the ?gure. Theoretical estimations for QDIPs are carried out assuming low quantum ef?ciency ≈2% (often mea-sured in practice) and 67%. The last value is typical for HgCdTe photodiodes (without antire?ection coat-ing). It should be noticed, however, that rapid progress has recently been made in the performance of QDIP devices, especially at near room temperature. Lim et al. have announced a quantum ef?ciency of 35% for detectors with peak detection wavelength around 4.1 μm [34].
Estimation of detectivity for InAs/GaInSb strained layer superlattices (SLSs) are based on several theoretical papers published previously [35–39]. Early calculations showed that a LWIR type-II InAs/GaInSb SLS should have an absorption coef?cient comparable to an HgCdTe alloy with the same cutoff wavelength [35]. Figure 7b predicts that type-II superlattices are the most ef?cient detector of IR radiation in long wavelength region. It is even better material than HgCdTe; it is characterized by high absorption coef?cient and relatively low thermal genera-tion rate. However, hitherto, this theoretical prediction has been not con?rmed by experimental data. The main rea-son of that is in?uence of Schockley-Read generation-recombination mechanism, which causes lower carrier life-time (higher thermal generation rate). It is clear from this analysis that the fundamental performance limitation of QWIPs is unlikely to rival HgCdTe photodetectors. However, the performance of very uniform QDIP [when σens /σQD = 1, see Eq. (6)] is predicted to rival with HgCdTe. We can also notice from Fig. 7 that GaAs/AlGaAs QWIP is better material than extrinsic silicon.
3.4. BLIP Temperature
BLIP temperature is de?ned that the device is operating at a temperature at which the dark current is equal to the background photocurrent, given a ?eld of view (FOV) and a background temperature.
In Fig. 8a, plots of the calculated temperature required for background limited operation in f /2 FOV are shown as a function of cutoff wavelength for various types of detectors. We can see that the operating temperature of QDIPs is comparable with HgCdTe photodiodes. HgCdTe detectors with background limited performance operate in practice with thermoelectric coolers in the MWIR range, but the LWIR detectors (8 ≤ λc ≤ 12 μm) operate at
η
qh ν
2
G th
------------------------.
D e t e c t i v i t y , c m H z 1/2W –1
Temperature, K
1024
10221020101810161014
1012
1010
D e t e c t i v i t y , c m H z 1/2W –1
Temperature, K
10101010101010(a)
(b)
Fig. 7. The predicted thermal detectivity versus temperature for various MWIR (λc = 5 μm) (a) and LWIR (λc = 5 μm) (b)photodetectors. The assumed quantum ef?ciencies are indicated.
242
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
≈100 K. HgCdTe photodiodes exhibit higher operating temperature compared to extrinsic detectors, silicide Schot-tky barriers, QWIPs and HTSCs. Type II SLSs are omitted in our considerations. The cooling requirements for QWIPs with cutoff wavelengths below 10 μm are less stringent in comparison with extrinsic detectors, Schottky barrier devices, and HTSCs.
Figure 8b gives additional insight on in?uence of QD nonuniformity on BLIP temperature. The quantum ef?-ciency for QWIPs, equal to 50%, has been assumed since QWIP cannot detect normal incidence radiation. It has been shown by Phillips [16] that the detector performance may be degraded by orders of magnitude for the values of σens /σQD = 100, which are indicative of the current state of QD fabrication technology. It is well known, that reduced optical absorption in QDs due to size nonuniformity results in an increase in the normalized dark current and a reduction in detectivity. The nonuniformity has also strong in?uence on BLIP temperature. Increase of σens /σQD ratio from 1 to 100 causes decrease of T BLIP by several tens of degrees.
4. EXPERIMENTAL VERIFICATION
4.1. Performance at Low Temperature
4.1.1. Dark current and RoA product. In spite of that QDIP is a photoconductor and HgCdTe photodiode, it is interesting to compare their dark currents and incremental resistances. At present stage of technology develop-ment, the dark currents of both detectors in the region of low bias voltages are comparable. Figure 9 displays the dependence of dark currents of 70-layer QDIP [40] and P-on-n HgCdTe photodiode [41] with a peak wavelength of 5 μm. The additional three calculated curves for HgCdTe photodiodes are derived from equation R 0A = kT /qJ s using measured R 0A values from Rockwell Scienti?c.
An additional insight in comparison of both types of devices gives Fig. 10, where the dependence of R 0A prod-uct on a wavelength is shown. The QDIP data was determined from dynamic resistance in I–V characteristics at operating bias. Only limited experimental R 0A values for QDIPs marked in Fig. 10 are available in literature. The highest measured R 0A values for HgCdTe photodiodes operated at 78 K with about 5-μm cutoff wavelength are located between 108 and 109 Ohm cm 2. The solid line is theoretical R 0A for HgCdTe photodiodes, calculated using a 1-D model that assumes diffusion current from narrower band gap n-side is dominant, and minority carrier recombination via Auger and radiative process. Theoretical calculations used typical values for the p-side donor concentration (N d = 1 × 1015 cm –3) and the narrow bandgap active layer thickness (10 μm). The R 0A product is inherent property of the HgCdTe ternary alloy and depends on cutoff wavelength. Dark current of photodiodes increases with cutoff wavelength, what is an important difference with QDIPs, where dark current is far less sen-
sitive to wavelength and depends on device geometry.
Cutoff wavelength, μm
T B L I P , K 24020016012080400
(a)
(b)
Cutoff wavelength, μm
240
2001601208040
Fig. 8. Estimation of the temperature required for background limited operation of: (a) various types of photodetectors, (b)QDIPs with different levels of quantum dot size nonuniformity.
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
243
4.1.2. Detectivity at 78 K. A useful ?gure of merit, for comparing detector performance, is thermally limited detectivity. Figure 11 compares the highest measurable detectivities at 77 K of QDIPs found in literature with the predicted detectivities of P-on-n HgCdTe and type II InAs/GaInSb SLS photodiodes. The solid lines are theoretical thermal limited detectivities for HgCdTe photodiodes, calculated using a 1-D model that assumes diffusion current from narrower band gap n-side is dominant, and minority carrier recombination via Auger and radiative process.In calculations of typical values for the p-side donor concentration (N d = 1 × 1015 cm –3), the narrow bandgap active layer thickness (10 μm), and quantum ef?ciency (60%) have been used. It should be insisted, that for HgCdTe pho-todiodes, theoretically predicted curves for temperature range between 50 and 100 K coincide very well with experimental data (not shown in Fig. 11). The predicted thermally limited detectivities of the type II SLS are larger than those for HgCdTe [39].
λ ≈ 5 μm
10–8
10–910–10
10–11
10–1210–13
Bias voltage, V
107 Ω cm 2
Exper. Data
70-layer QDIP (at 77 K)
HgCdTe photodiode (at 98 K)
108 Ω cm 2
109 Ω cm 2
–0.5
–0.4
–0.3
–0.2
–0.10Fig. 9. Comparison of dark currents of 70-layer QDIP (after [40]) and P-on-n HgCdTe photodiode (after [41]) with a peak wavelength of 5 μm. The additional three calculated curves for HgCdTe photodiodes are derived from equation R 0A = kT /qJ s using measured R 0A values from Rockwell Scienti?c (after [42]).
R 0A , Ω c m 2
Wavelength, μm
Experimental data at 78 K
70-layer ODIPs [43]P-on-n HgCdTe [42]
HgCdTe (theory)
10114
6
8
10
12
1416182010910710510310110–110–3
Fig. 10. R 0A vs. wavelength for P-on-n HgCdTe photodiodes and QDIPs at 78 K. Solid line is calculated theoretically assuming 1-D n-side diffusion model.
D a r k c u r r e n t d e n s i t y , A /c m 2
244
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
The measured value of QDIPs’ detectivities at 77 K gathered in Fig. 11 indicate that QD device detectivities are as yet considerably inferior to current HgCdTe detector performance. In LWIR region, the upper experimental QDIP data at 77 K coincide with HgCdTe ones at temperature 100 K.
4.2. Performance at Higher Temperature
4.2.1. Dark current and R 0A product. One of the main potential advantages of QDIPs is low dark current. In particular, the lower dark currents enable higher operating temperatures. Up till now, however, most of the QDIP devices reported in the literature have been working in the temperature range of 77 to 200 K. On account of this fact, it is interesting to insight on achievable QDIP performance in temperature range above 200 K in comparison with other type of detectors.
Most modern infrared devices are fabricated from two pieces of material—a detector array made from com-pound semiconductor materials and a silicon signal processing chip called a readout integrated circuit (ROIC). The ROIC ampli?ers the signal from each detector element and performs processing functions by multiplexing the sig-nals of thousands of pixels onto a few output lines.
The infrared arrays have individual-ampli?er-per-detector readouts based on metal oxide semiconductor ?eld effect transistors (MOSFETs). The operating point of the coupled detector and input circuit is found by construct-ing a load line for the I–V characteristics of the detector and input MOSFET. The input impedance of a MOSFET is a function of the source-drain current (in this case, the total diode current) and is usually expressed in terms of the transconductance, g m , given by qI /(nkT ) for low injected currents (n is an ideality factor that can vary with tem-perature and geometry of the transistor and usually is in the range 1–2).The injection ef?ciency is approximately given by
(13)
where R d is the dynamic impedance of the detector and I is the total injected detector current (the sum of the photocurrent and the dark current) equal photocurrent, I ph , in the background-limited case.
To receive high injection ef?ciency, the input impedance of the MOSFET must be much lower than the internal dynamic resistance of the detector at its operating point, and the following condition should be ful?lled [52]
(14)
For most applications, the detector performance depends on operating the detector in a small bias where the dynamic resistance is at a maximum. It is then necessary to minimize extraneous leakage current. The control of these leakage currents and the associated low-frequency noise is therefore of crucial interest.
Generally, is not problem to ful?l this inequality for short wavelength infrared (SWIR) and middle wavelength infrared (MWIR) focal plane arrays (FPAs) where the dynamic resistance of detector R d , is large, but it is very
εIR
d
IR d nkT /q ()
+----------------------------------.=IR d ? nkT q
---------.
D e t e c t i v i t y , c m H z 1/2/W Wavelength, μm
Type II SLS (theory)
QDIPs (experimental data at 77 K)HgCdTe ((theory)
40 K 78 K
40 K 50 K 65 K 78 K 100 K
[34][44][45][46][47][48][49][50][51]
1017101510131011109107
4
6
8
101214
161820
Fig. 11. The predicted detectivity of P-on-n HgCdTe and type II InAs/GaInSb SLS photodiodes, compared with measured QDIP detectivities at 77 K.
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
245
important for LWIR designs where R d is low. There are more complex injection circuits that effectively reduce the input impedance and allow lower detector resistance to be used.
The above requirement is especially critical for near-room temperature HgCdTe photodetectors operating in LWIR region. Their resistance is very low due to a high thermal generation. In materials with a high electron to hole ration as HgCdTe, the resistance is additionally reduced by ambipolar effects. Small size uncooled 10.6-μm photodiodes (50 × 50 μm 2) exhibit less than 1 Ω zero bias junction resistances which are well below the series resis-tance of a diode. As a result, the performance of conventional devices is very poor, so they are not usable for prac-tical applications. To ful?l inequality (14) to effectively couple the detector with silicon readout, the detector incre-mental resistance should be R d ? 2Ω. As Fig. 12 shows, the saturation current for 10-μm photodiode achieves 1000 A/cm 2 and it is by four orders of magnitude larger than the photocurrent due to the 300 K background radia-J s , A /c m 2
Cutoff wavelength, μm
HgCdTe photodiodes T = 300 K
T = 250 K T = 200 K
J B (300 K, FOV = 2π)QDIPs
105510
1520103101
10–110–310–5
Fig. 12. Dark current density of HgCdTe photodiodes and QDIPs, and background-generated photocurrent as a function of wave-length. The calculations for HgCdTe photodiodes have been performed for the optimized doping concentration p = γ1/2n i .
R 0A , Ω
c m 2
Cutoff wavelength, μm
103
10110–110–310–510–7
Fig. 13. R 0A product of HgCdTe photodiodes and QDIPs as a function of wavelength. The calculations for HgCdTe photo-diodes have been performed for the optimized doping concentration p = γ1/2n i .
246
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
tion. The potential advantages of QDIPs is considerably lower dark current and higher R 0A product in comparison with HgCdTe photodiodes (see Fig. 13).
4.2.2. Detectivity at room temperature. Figure 14 compares the calculated thermal detectivity of HgCdTe photodiodes and QDIPs as a function of wavelength and operating temperature with the experimental data of uncooled HgCdTe and type-II InAs/GaInSb SLS detectors. The Auger mechanism is likely to impose fundamental limitations to the LWIR HgCdTe detector performance. The calculations have been performed for optimized dop-ing concentration p = γ1/2n i . The experimental data for QDIPs are gathered from the literature for detectors operated at 200 and 300 K.
Uncooled LWIR HgCdTe photodetectors are commercially available and manufactured in signi?cant quanti-ties, mostly as single-element devices [53]. They have found important applications in IR systems that require fast response. The results presented in Fig. 14 con?rm that the type-II superlattice is a good candidate for IR detectors operating in the spectral range from the mid-wavelength to the very long-wavelength IR. However, comparison of QDIP performance both with HgCdTe and type II superlattice detectors gives clear evidence that the QDIP is suit-able for high temperature. Especially encouraging results have been achieved for very long-wavelength QDIP devices with a double-barrier resonant tunnelling ?lter with each quantum-dot layer in the absorption region [54, 55]. In this type of devices, photoelectrons are selectively collected from the QDs by resonant tunnelling, while the same tunnel barriers block electrons of dark current due to their broad energy distribution. For the 17-μm detec-tor, a peak detectivity of 8.5 × 106 cm Hz 1/2/W has been measured. Up till now, this novel device demonstrates the highest performance of room-temperature photodetectors. Further improvement in technology and design can result in application of QDIPs in room temperature focal plane arrays with the advantages of larger operating speed (shorter frame time) in comparison with thermal detectors (bolometers and pyroelectric devices).
The room-temperature operation of thermal detectors makes them lightweight, rugged, reliable, and convenient to use. However, their performance is modest, and they suffer from slow response. Because they are nonselective detectors, their imaging systems contain very broadband optics, which provide impressive sensitivity at a short
range in good atmospheres.
D *, c m H z 1/2/W
Cutoff wavelength, μm
Uncooled:
10
101010109
108107
Fig. 14. Calculated performance of Auger generation-recombination limited HgCdTe photodetectors as a function of wavelength and operating temperature. BLIP detectivity has been calculated for 2π FOV , the background temperature is T BLIP = 300 K, and the quantum ef?ciency η = 1. The calculations for HgCdTr photodiodes have been performed for the optimized doping concen-tration p = γ1/2n i . The experimental data is taken for commercially available uncooled HgCdTe photoconductors (produced by Vigo System) and uncooled type-II detectors at the Center for Quantum Devices, Northwestern University (U.S.). The experimen-tal data for QDIPs are gathered from the marked literature for detectors operated at 200 and 300 K.
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
247
Thermal detectors seem to be unsuitable for the next generation of IR thermal imaging systems, which are mov-ing toward faster frame rates and multispectral operation. A response time much shorter than that achievable with thermal detectors is required for many nonimaging applications. Improvement in technology and design of QDIP detectors make it possible to achieve both high sensitivity and fast response at room temperature.
5. FOCAL PLANE ARRAYS 5.1. Single-Colour Devices
In the past three years, several groups have independently demonstrated cryogenically cooled quantum dot FPAs with midformat size: 256 × 256, [57, 58] 320 × 256, [59] and 640 × 512 [11, 21].
Recently, the dot-in-well (DWELL) structures with peak responsivity out to 8.1 μm and detectivity ≈1010 cm Hz 1/2/W at 77 K fabricated into 640 × 512 pixel FPA produced infrared imagery with NEDT of 40 mK at 60-K operating temperature [21]. Figure 15a shows detectivity of this detector as a function of temperature at operating bias –1 V . This ?gure shows that the DWELL detector reaches background limited D * about 1011 Jones around 50 K. Next ?gure (Fig. 15b) shows an image taken with this LW QDIP FPAs. It should be noticed that the performance of DWELL-QDIP presented here still lags behind that for the state-of-the-art QWIPs.
5.2. Multi-Colour Devices
In the IR wavelength region two detector technologies that are developing multicolour capability are well estab-lished; HgCdTe photodiodes and quantum well infrared photodetectors (QWIPs). Both HgCdTe photodiodes and QWIPs offer the multicolour capability in the SWIR, MWIR and LWIR range. The performance ?gures of merit of state-of-the-art QWIP and HgCdTe FPAs are similar because the main limitations come from the readout cir-cuits. A more detailed comparison of both technologies has been given by Tidrow et al. [60] and Rogalski [15].Recently, type II InAs/GaInSb superlattices [61–63] and QDIPs [51, 59, 64–66] have emerged as next two can-didates for third generation infrared detectors.
QDIP devices capable of detecting several separate wavelengths can be fabricated by vertical stacking of the different QWIP layers during epitaxial growth. The schematic structure is shown in Fig. 16. In the case of structure describing by Lu et al. [66] each of QDIP absorption band consists of 10-period of InAs/InGaAs QD layers sand-wiched between the top and bottom electrodes. Next ?gure shows the simpli?ed band diagram of this structure at different bias levels. As Fig. 17 presents, the bias voltage selection of detection bands originates from the asymmet-ric band structure. At low bias voltage, the high energy GaAs barrier blocks the photocurrent generated by LWIR radiation and only responses to the MWIR incidence. On the contrary, as the bias voltage increases, the barrier
energy decreases, allowing LWIR signals to be detected at different bias voltage levels.
D e t e c t i v i t y , c m H z 1/2W –1
Temperature, K
V b = –1 V
101140
50607080
90100(a)
(b)
1010109108Fig. 15. 640 × 512 DWELL-QDIP FPAs (300 K, f/2 optics): (a) detectvity as a function of operating temperature, (b) image taken with FPA camera (after [21]).
248OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
The ?rst two-colour quantum dot FPA demonstration was based on a voltage-tuneable InAs/InGaAs/GaAs DWELL structure [51, 59]. The InAs QDs are placed in an InGaAs well, which in turn is placed in a GaAs matrix (see Fig. 14).
Figure 18 shows a multicolour response from the DWELL detectors. This device has demonstrated multicolour response ranging from the MWIR (3–5 μm) based on a bound-to-continuum transition to the LWIR (8–12 μm)based on a bound state in the dot to a bound state in the well. A very long wavelength response (VLWIR) has also been observed and has been attributed to transitions between two bound states in the QDs since the calculated energy spacing between the dot levels is about 50–60 meV . Moreover, by adjusting the voltage bias on the device,one can change the ratio of the electrons promoted by MWIR, LWIR, and VLWIR absorptions. Typically, the
Tunneling
Blocked
GaAs (a)
(b)
InGaAs barriers
InAs QDs
MWIR LWIR
barriers
InAs QDs
Tunneling
GaAs InGaAs barriers
InAs QDs
MWIR LWIR barriers
InAs QDs
Tunneling Fig. 17.
Simpli?ed band diagram of the structure shown in Fig. 16 at low (a) and higher (b) bias levels (after [66]).
n* GaAs contacting layer
GaAs InAs QDs InGaAs MWIR LWIR
Semiinsulating GaAs substrate
InAs QDs GaAs GaAs GaAs n* GaAs contacting layer x10
x10Fig. 16. Schematic structure of the multispectral QDIP device (after [66]).
R e s p o n s i v i t y , V /W
0.0300.0250.0200.0150.0100.005
Wavelength, μm
Fig. 18. Multicolour response from an InAs/In 0.15Ga 0.85As/GaAs DWEL detector. The MWIR (LWIR) peak is possibly a transition from a state in the dot to a higher (lower) lying state in the well whereas the VLWIR response is possibly from two quantum-con?ned levels within the QD. This response is visible till 80 K (after [51]).
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS
249
MWIR response dominates at low to nominal voltages due to higher escape probability. With the voltage increas-ing, the LWIR and eventually VLWIR responses enhance due to the increased tunnelling probability of lower lying states in the DWELL detector. Bias-dependent shift of the spectral response is observed due to quantum-con?ned Stark effect. This spectral diversity can be exploited to realize spectrally smart sensors whose wavelength and band-width can be tuned depending on the desired application [51, 67].
Typically, the detector structure consists of a 15-stack asymmetric DWELL structure sandwiched between two highly doped n-GaAs contact layers. The DWELL region consists of a 2.2 ML of n-doped InAs QDs in an In 0.15Ga 0.85As well, itself placed in GaAs. By varying the width of the bottom InGaAs well from 10 to 60 ?, the operating wavelength of the detector was changed from 7.2 to 11 μm. The responsivity and detectivity obtained from the test devices at 78 K are shown in Fig. 19. The measured detectivities were of 2.6 × 1010 cm Hz 1/2/W (V b =2.6 V) for the LWIR band, and 7.1 × 1010 cm Hz 1/2/W (V b = 1 V) for the MWIR band.
Recently, V arley et al. [65] have demonstrated a two-colour, MWIR/LWIR, 320 × 256 FPA based on DWELL detectors. Minimum NEDT
values of 55 mK (MWIR) and 70 mK (LWIR) were measured (see Fig. 20).
R e s p o n s i v i t y , A /W
Bias voltage, V
D e t e c t i v i t y , J o n e s
101
8 × 10106 × 10104 × 10102 × 1010
10010–110–210–310–4
Fig. 19. Peak responsivity for a 15 stack DWELL detector at 78 K obtained using a calibrated blackbody source. Solid squares: MWIR responsivity; solid triangles: LWIR responsivity; open square: MWIR detectivity; open triangles: LWIR detectivity (after [51]).
N E D T , K
Irradiance, photons/cm 2 s
MWIR LWIR
0.30
0.20
0.10
01015
1016
10171018
Fig. 20. NEDT in the MWIR and LWIR bands at 77 K. Irradiance levels for MWIR and LWIR are 3–5 μm (f/2) and 8–12 μm (f /2.3), respectively (after [65]).
250OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
6. CONCLUSIONS
The paper compares the achievements of QDIP technology with those of competitive technologies, with the emphasis on the material properties, device structure, and their impact on the device performance, especially in LWIR spectral regions.
At present, HgCdTe is the most widely used variable-gap semiconductor that has a privileged position both in the MWIR as well as LWIR spectral ranges. Theoretical predictions indicate that only type II superlattice photo-diodes and QDIPs are expected to compete with HgCdTe photodiodes. However, the measured values of QDIPs’detectivities at 77 K are considerably inferior to current HgCdTe detector performance. Improving QD uniformity is a key issue in the increasing the absorption coef?cient and improving the performance.
Comparison of QDIP performance both with HgCdTe and type II superlattice detectors gave evidence that the QDIP is suitable for high operation temperature. Especially encouraging results have been recently achieved for very long-wavelength QDIP devices with a double-barrier resonant tunnelling ?lter. Due to fact that conventional HgCdTe photodiodes are not usable for room temperature FPA applications, it can be expected that improvement in technology and design of QDIP detectors will make it possible to achieve both high sensitivity and fast response useful for practical application in room temperature FPAs. This new generation of room temperature FPAs will eventually compete with silicon microbolometers, dominant at present. Larger operating speed of QDIP and mul-tispectral capability are considerable advantages in comparison with thermal detectors.
Optimization of the QDIP architecture is still an open area. Since some of the design parameters depend on a device structure (photoconductive and noise gains, dark current, quantum ef?ciency), the performance is still being improved.
QDIP detector technology is on very early stage of development. The bias-dependent spectral diversity of this type of detectors can be exploited to realize multispectrally smart sensors whose wavelength and bandwidth can be tuned depending on the desired application. Optimization of the QDIP architecture is still an open area.
The thickness of the entire vertical pixel structure of type-II photodiodes and QDIPs is only about 5 μm, which signi?cantly reduces the technological challenge compared to bispectral HgCdTe FPAs. Bulk two-colour detec-tors, with a typical total layer thickness above 10 μm, require deeper isolation trenches—at least 15 μm deep trenches and no more than 5 μm wide at the top for a such small as 20 μm pitch devices.
REFERENCES
1.Esaki, L. and Tsu, R., Superlattice and Negative Conductivity in Semiconductors, IBM J. Res. Develop., 1970,vol. 14, pp. 61–65.
2.Arakawa, Y . and Sakaki, H., Multidimensional Quantum-Well Laser and Temperature Dependence of Its Thresh-old Current, Appl. Phys. Lett., 1982, vol. 40, pp. 939–941.
3.Asada, M., Miyamoto, Y ., and Suematsu, Y ., Gain and Threshold of Three Dimensional Quantum-Box Lasers,IEEE J. Quant. Electron., 1986, vol. 22, pp. 1915–1921.
4.Bimberg, D., Grundmann, M., and Ledentsov, N.N., Quantum Dot Heterostructures , Chichester: Wiley, 1999.
5.Kinch, M.A., Fundamental Physics of Infrared Detector Materials, J. Electron. Mater., 2000, vol. 29, pp. 809–817.
6.Sikorski, Ch. and Merkt, U., Spectroscopy of Electronic States in InSb Quantum Dots, Phys. Rev. Lett., 1989,vol. 62, pp. 2164–216
7.
7.Demel, T., Heitmann, D., Grambow, P., and Ploog, K., Nonlocal Dynamic Response and Level Crossings in Quan-tum-Dot Structures, Phys. Rev. Lett., 1990, vol. 64, pp. 788–791.
8.Phillips, J., Kamath, K., and Bhattacharya, P., Far-Infrared Photoconductivity in Self-Organized InAs Quantum Wells, Appl. Phys. Lett., 1998, vol. 72, pp. 2020–2021.
9.Bhattacharya, P. and Mi, Z., Quantum-Dot Optoelectronic Devices, Proc. IEEE , 2007, vol. 95, pp. 1723–1740.10.Campbell, J.C. and Madhukar, A., Quantum-Dot Infrared Photodetectors, Proc. IEEE , 2007, vol. 95,
pp. 1815?1827.
11.Krishna, S., Gunapala, S.D., Bandara, S.V ., Hill, C., and Ting, D.Z., Quantum Dot Based Infrared Focal Plane
Arrays, Proc. IEEE , 2007, vol. 95, pp. 1838–1852.
12.Leonard, D., Krishnamurthy, M., Reaves, C.M., Denbaars, S.P., and Petroff, P.M., Direct Formation of Quantum-Sized Dots from Uniform Coherent Islands of InGaAs on GaAs Surface, Appl. Phys. Lett., 1993, vol. 63,pp. 3203?3205.
13.Stranski, I.N. and Krastanow, L., Zur theorie der orientierten ausscheidung von lonenkristallen aufeinander, Sit-zungsberichte d. Akad. d. Wissenschaften in Wein. Abt. IIb , 1937, vol. 146, pp. 797–810.
14.Gunapala, S.D. and Bandara, S.V GaAs/AlGaAs Based Quantum Well Infrared Photodetector Focal Plane Arrays, in
Handbook of Infrared Detection Technologies , Henini, M. and Razeghi, M., Ed., Oxford: Elsevier, 2002, pp. 83–119.15.Rogalski, A., Quantum Well Photoconductors in Infrared Detector Technology, J. Appl. Phys., 2003, vol. 93,
pp. 4355–4391.
OUTLOOK ON QUANTUM DOT INFRARED PHOTODETECTORS251 16.Phillips, J., Evaluation of the Fundamental Properties of Quantum Dot Infrared Detectors, J. Appl. Phys., 2002,
vol. 91, pp. 4590–4594.
17.Liu, H.C., Quantum Dot Infrared Photodetector, Opto-Electron. Rev., 2003, vol. 11, pp. 1–5 (2003).
18.Wang, S.Y., Lin, S.D., Wu, W., and Lee, C.P., Low Dark Current Quantum-Dot Infrared Photodetectors with an
AlGaAs Current Blocking Layer, Appl. Phys. Lett., 2001, vol. 78, pp. 1023–1025.
19.Ryzhii, V., Physical Model and Analysis of Quantum Dot Infrared Photodetectors with Blocking Layer, J. Appl.
Phys., 2001, vol. 89, pp. 5117–5124.
20.Krishna, S., Quantum Dots-in-a-Well Infrared Photodetectors, J. Phys., Ser. D: Appl. Phys., 2005, vol. 38,
pp. 2142–2150.
21.Gunapala, S.D., Bandara, S.V., Hill, C.J., Ting, D.Z., Liu, J.K., Rafol, B., Blazejewski, E.R., Mumolo, J.M., Keo,
S.A., Krishna, S., Chang, Y.-C., and Shott, C.A., 640 × 512 Pixels Long-Wavelength Infrared (LWIR) Quantum-Dot Infrared Photoconductor (QDIP) Imaging Focal Plane Array, IEEE J. Quantum Electron., 2007, vol. 43, pp. 230–237.
22.Lee, S.W., Hirakawa, K., and Shimada, Y., Bound-to-Continuum Intersubband Photoconductivity of Self-Assem-
bled InAs Quantum Dots in Modulation-Doped Heterostructures, Appl. Phys. Lett., 1999, vol. 75, pp. 1428–1430.
23.Vurgaftman, I., Lam, Y., and Singh, J., Carrier Thermalization in Sub-Three-Dimensional Electronic Systems:
Fundamental Limits on Modulation Bandwidth in Semiconductor Lasers, Phys. Rev., Ser. B, 1994, vol. 50, pp. 14309–14326.
24.Towe, E. and Pan, D., Semiconductor Quantum-Dot Nanostructures: Their Application in a New Class of Infrared
Photodetectors, IEEE J. Selected Topics Quantum Electron., 2000, vol. 6, pp. 408–421.
25.Long, D., Photovoltaic and Photconductive Infrared Detectors, in Optical and Infrared Detectors, Keyes, R.J., Ed.,
Berlin: Springer-Verlag, 1977, pp. 101–147.
26.Elliott, C.T. and Gordon, N.T., Infrared Detectors, in Handbook on Semiconductors, Hilsum, C., Ed., North-Hol-
land, Amsterdam, 1993, vol. 4, pp. 841–936.
27.Elliott, C.T. and Jones, C.L., Non-Equilibrium Devices in HgCdTe, in Narrow-Gap II–VI Compounds for Opto-
electronic and Electromagnetic Applications, Capper, P., Ed., London: Chapman & Hall, 1997, pp. 474–485. 28.Piotrowski, J. and Gawron, W., Ultimate Performance of Infrared Photodetectors and Figure of Merit of Detector
Material, Infrared Phys. Technol., 1997, vol. 38, pp. 63–68.
29.Singh, J., Electronic and Optoelectronic Properties of Semiconductor Structures, Cambridge: Cambridge Univer-
sity Press, 2003.
30.Brounkov, P.N., Polimeni, A., Stoddart, S.T., Henini, M., Eaves, L., Main, P.C., Kovsh, A.R., Musikhin, Yu.G., and
Konnikov, S.G., Electronic Structure of Self-Assembled InAs Quantum Dots in GaAs Matrix, Appl. Phys. Lett., 1998, vol. 73, pp. 1092–1094.
31.Jiang, H. and Singh, J., Self-Assembled Semiconductor Structures: Electronic and Optoelectronic Properties,
IEEE J. Quantum Electron., 1998, vol. 34, pp. 1188–1196.
32.Rogalski, A., Adamiec, K., and Rutkowski, J., Narrow-Gap Semiconductor Photodiodes, Bellingham: SPIE Press,
2000.
33.Rogalski, A., HgCdTe Infrared Detector Material: History, Status and Outlook, Rep. Prog. Phys., 2005, vol. 68,
pp. 2267–2336.
34.Lim, H., Tsao, S., Zhang, W., and Razeghi, M., High-Performance InAs Quantum-Dot Infrared Photoconductors
Grown on InP Substrate Operating at Room Temperature, Appl. Phys. Lett., 2007, vol. 90, p. 131112.
35.Smith, D.L. and Mailhiot, C., Proposal for Strained Type II Superlattice Infrared Detectors, J. Appl. Phys., 1987,
vol. 62, pp. 2545–2548.
36.Grein, C.H., Young, P.M., and Ehrenreich, H., Minority Carrier Lifetimes in Ideal InGaSb/InAs Superlattice, Appl.
Phys. Lett., 1992, vol. 61, pp. 2905–2907.
37.Youngdale, E.R., Meyer, J.R., Hoffman, C.A., Bartoli, F.J., Grein, C.H., Young, P.M., Ehrenreich, H., Miles, R.H.,
and Chow, D.H., Auger Lifetime Enhancement in InAs–Ga1–xIn x Sb Superlattices, Appl. Phys. Lett., 1994, vol. 64, pp. 3160–3162.
38.Grein, C.H., Cruz, H., Flatte, M.E., and Ehrenreich, H., Theoretical Performance of Very Long Wavelength
InAs/In x Ga1 – x Sb Superlattice Based Infrared Detectors, Appl. Phys. Lett., 1994, vol. 65, pp. 2530–2532.
39.Grein, C.H., Young, P.M., Flatté, M.E., and Ehrenreich, H., Long Wavelength InAs/InGaSb Infrared Detectors:
Optimization of Carrier Lifetimes, J. Appl. Phys., 1995, vol. 78, pp. 7143–7152.
40.Bhattacharya, P., Chakrabarti, S., Su, X., and Stiff-Roberts, A.D., Research Propels Quantum Dots Forward, Laser
Focus World, 2005, vol. 41, pp. 103–108.
41.Wijewarnasuriya, P.S., Zandian, M., Phillips, J., Edwall, D., DeWames, R.E., Hildebrandt, G., Bajaj, J., Arias, J.M.,
D’Souza, A.I., and Moore, F., Advances in Large-Area Hg1–x Cd x Te Photovoltaic Detectors for Remote-Sensing Applications, J. Electron. Mater., 2002, vol. 31, pp. 726–731.
42.Chuh, T., Recent Developments in Infrared and Visible Imaging for Astronomy, Defense and Homeland Security,
Proc. SPIE, 2004, vol. 5563, pp. 19–34.
OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS)V ol. 18No. 32009
252OPTICAL MEMORY AND NEURAL NETWORKS (INFORMATION OPTICS) V ol. 18 No. 3 2009
ROGALSKI
43.Bhattacharya, P., Stiff-Roberts, A.D., and Chakrabarti, S., Mid-Infrared Quantum Dot Photoconductors, in Mid-Infrared Semiconductor Optoelectronics , Krier, A., Ed., Springer Verlag, 2007, pp. 487–513.
44.Jiang, J., Tsao, S., O’Sullivan, T., Zhang, W., Lim, H., Sills, T., Mi, K., Razeghi, M., Brown, G.J., and Tidrow, M.Z.,
High Detectivity InGaAs/InGaP Quantum-Dot Infrared Photodetectors Grown by Low Pressure Metalorganic Chemical Vapor Deposition, Appl. Phys. Lett., 2004, vol. 84, pp. 2166–2168.
45.Szafraniec, J., Tsao, S., Zhang, W., Lim, H., Taguchi, M., Quivy, A.A., Movaghar, B., and Razeghi, M., High-Detectivity Quantum-Dot Infrared Photodetectors Grown by Metalorganic Chemical-Vapor Deposition, Appl.Phys. Lett., 2006, vol. 88, p. 121102.
46.Kim, E.-T., Madhukar, A., Ye, Z., and Campbell, J.C., High Detectivity InAs Quantum Dot Infrared Photodetec-tors, Appl. Phys. Lett., 2004, vol. 84, pp. 3277–3279.
47.Gunapala, S.D., Bandara, S.V ., Hill, C.J., Ting, D.Z., Liu, J.K., Rafol, B., Blazejewski, E.R., Mumolo, J.M., Keo,
S.A., Krishna, S., Chang, Y .-C., and Shott, C.A., Long-Wavelength Infrared (LWIR) Quantum Dot Infrared Pho-todetector (QDIP) Focal Plane Array, Proc. SPIE , 2006, vol. 6206, p. 62060J.
48.Chakrabarti, S., Su, X.H., Bhattacharya, P., Ariyawansa, G., and Perera, A.G.U., Characteristics of a Multicolor
InGaAs-GaAs Quantum-Dot Infrared Photodetector, IEEE Phot. Technol. Lett., 2005, vol. 17, pp. 178–180.
49.Attaluri, R.S., Annamalai, S., Posani, K.T., Stintz, A., and Krishna, S., In?uence of Si Doping on the Performance
of Quantum Dots-in-Well Photodetectors, J. Vac. Sci. Technol., Ser. B, 2006, vol. 24, pp. 1553–1555.
50.Chakrabarti, S., Stiff-Roberts, A.D., Su, X.H., Bhttacharya, P., Ariyawansa, G., and Perera, A.G.U., High-Perfor-mance Mid-Infrared Quantum Dot Infrared Photodetectors, J. Phys., Ser. D: Appl. Phys., 2005, vol. 38,pp. 2135?2141.
51.Krishna, S., Forman, D., Annamalai, S., Dowd, P., Varangis, P., Tumolillo, T., Gray, A., Zilko, J., Sun, K., Liu, M.,
Campbell, J., and Carothers, D., Two-Color Focal Plane Arrays Based on Self Assembled Quantum Dots in a Well Heterostructure, Phys. Stat. Sol., Ser. C , 2006, vol. 3, pp. 439–443.
52.Vampola, J.L., Readout Electronic for Infrared Sensors, in The Infrared and Electro-Optical Systems Handbook ,
Rogatto, W.D., Ed., Bellingham: SPIE Optical Engineering Press, 1993, vol. 3, pp. 285–342.
53.Piotrowski, J. and Rogalski, A., High-Operating Temperature Infrared Photodetectors , Bellingham: SPIE Press,
2007.
54.Su, X.H., Chakrabarti, S., Bhattacharya, P., Ariyawansa, A., and Perera, A.G.U., A Resonant Tunneling Quantum-Dot Infrared Photodetector, IEEE J. Quant. Electron., 2005, vol. 41, pp. 974–979.
55.Perera, A.G.U., Ariyawansa, G., Apalkov, V .M., Matsik, S.G., Su, X.H., Chakrabarti, S., and Bhattacharya, P.,
Wavelength and Polarization Selective Multi-Band Tunnelling Quantum Dot Detectors, Opto-Electron. Rev.,2007, vol. 15, pp. 223–228.
56.Lu, X., Vaillancourt, J., and Meisner, M.J., Temperature-Dependent Photoresponsivity and High-Temperature
(190 K) Operation of a Quantum Dot Infrared Photodetector, Appl. Phys. Lett., 2007, vol. 91, p. 051115.
57.Jiang, J., Mi, K., Tsao, S., Zhang, W., Lim, H., O’Sullivan, T., Sills, T., Razeghi, M., Brown, G.J., and Tidrow,
M.Z., Demonstration of a 256 × 256 Middle-Wavelength Infrared Focal Plane Array Based on GaAs/InGaP Quan-tum Dot Infrared Photodetectors, Appl. Phys. Lett., 2004, vol. 84, pp. 2232–2234.
58.Tang, S.-F., Chiang, C.-D., Weng, P.-K., Gau, Y .-T., Luo, J.-J., Yang, S.-T., Shih, C.-C., Lin, A.-Y ., and Lee, S.-C.,
High-Temperature Operation Normal Incident 256 × 256 InAs-GaAs Quantum-Dot Infrared Photodetector Focal Plane Array, IEEE Photon. Technol. Lett., 2006, vol. 18, pp. 986–988.
59.Krishna, S., Forman, D., Annamalai, S., Dowd, P., Varangis, P., Tumolillo, T., Gray, A., Zilko, J., Sun, K., Liu, M.,
Campbell, J., and Carothers, D., Demonstration of a 320 × 256 Two-Color Focal Plane Array Using InAs/InGaAs Quantum Dots in Well Detectors, Appl. Phys. Lett., 2005, vol. 86, p. 193501.
60.Tidrow, M.Z., Beck, W.A., Clark, W.W., Pollehn, H.K., Little, J.W., Dhar, N.K., Leavitt, P.R., Kennedy, S.W.,
Beekman, D.W., Goldberg, A.C., and Dyer, W.R., Device Physics and Focal Plane Applications of QWIP and MCT, Opto-Electron. Rev., 1999, vol. 7, pp. 283–296.
61.Rehm, R., Walther, M., Schmitz, J., Flei?er, J., Fuchs, F., Ziegler, J., and Cabanski, W., InAs/GaSb Superlattice
Focal Plane Arrays for High-Resolution Thermal Imaging, Opto-Electron. Rev., 2006, vol. 14, pp. 283–296.
62.Rehm, R., Walther, M., Schmitz, J., Flei?ner, J., Ziegler, J., Cabanski, W., and Breiter, R., 2nd and 3rd Generation
Thermal Images Based on Type-II Superlattice Photodiodes, Proc. SPIE , 2006, vol. 6294, p. 629404.
63.Rogalski, A. and Martyniuk, P., InAs/GaInSb Superlattices as a Promising Material System for Third Generation
Infrared Detectors, Infrared Phys. Technol., 2006, vol. 48, pp. 39–52.
64.Kim, S.M. and Harris, J.S., Multicolor InGaAs Quantum-Dot Infrared Photodetectors, IEEE Photon. Technol.
Lett., 2004, vol. 16, pp. 2538–2540.
65.Varley, E., Lenz, M., Lee, S.J., Brown, J.S., Ramirez, D.A., Stintz, A., and Krishna, S., Single Bump, Two-Color
Quantum Dot Camera, Appl. Phys. Lett., 2007, vol. 91, p. 081120.
66.Lu, X., Vaillancourt, J., and Meisner, M., A V oltage-Tunable Multiband Wuantum Dot Infrared Focal Plane Array
with High Photoconductivity, Proc. SPIE , 2007, vol. 6542, p. 65420Q.
67.Perera, A.G.U., Quantum Structures for Multiband Photon Detection, Opto-Electron. Rev., 2006, vol. 14,
pp. 99?108.